0.3.7 TTS Streaming regression - sentences are always buffered

[K-RAD]

Maybe I'm doing something wrong, but I can't get the Kokoro model (via speaches) to be streaming

[roryeckel]

Are you using STT_STREAMING_MODELS? How exactly could you tell streaming wasn't working? I don't see any logs, and really that's the easiest way to tell if it's working. It's not much of a noticeable improvement unless the hardware is weak

[K-RAD]

Are you using STT_STREAMING_MODELS? How exactly could you tell streaming wasn't working? I don't see any logs, and really that's the easiest way to tell if it's working. It's not much of a noticeable improvement unless the hardware is weak

I set it as:

      TTS_OPENAI_URL: http://speaches:8000/v1
      TTS_MODELS: "speaches-ai/Kokoro-82M-v1.0-ONNX"
      TTS_STREAMING_MODELS: "speaches-ai/Kokoro-82M-v1.0-ONNX"
      TTS_BACKEND: "SPEACHES"

I'll try to get some logs. But it's very obvious it's not streaming when I'm saying "tell me a short story" and it starts talking exactly after it finished writing. Hardware is not weak, running it on 16GB VRAM GPU + 64 GB RAM

[roryeckel]

The difference that you should be looking for is the time between when STT is completely processed, the initial time to start hearing a response is very slightly shorter. It doesn't change much beyond that, and it also depends on how long each sentence is in the response. If the response is one long sentence, then there won't be any chunks broken up to stream. Additionally, the openai library is using the streaming response here (even before the addition of TTS_STREAMING_MODELS)

[roryeckel]

Also, older versions of Home Assistant don't stream the transcription so they cannot stream tts very effectively, it still somewhat has an effect though

[K-RAD]

If I'm reading the logs correctly, it should be streaming. However the sound only starts after it finished writing everything..

2025-09-24T15:44:06.067136558Z DEBUG:wyoming_openai.handler:Incoming event type transcribe
2025-09-24T15:44:06.067202203Z DEBUG:wyoming_openai.handler:Incoming event type audio-start
2025-09-24T15:44:06.067297210Z INFO:wyoming_openai.handler:Recording started at 16000 Hz, 1 channels, 2 bytes per sample
2025-09-24T15:44:09.444395239Z DEBUG:wyoming_openai.handler:Incoming event type audio-stop
2025-09-24T15:44:09.444461978Z DEBUG:wyoming_openai.handler:Adding vad_filter=False for SPEACHES backend
2025-09-24T15:44:09.445544426Z DEBUG:openai._base_client:Request options: {'method': 'post', 'url': '/audio/transcriptions', 'headers': {'Content-Type': 'multipart/form-data'}, 'files': [('file', SerializationIterator(index=0, iterator=SerializationIterator(index=0, iterator=<wyoming_openai.utilities.NamedBytesIO object at 0x78efe7b8cf90>)))], 'idempotency_key': 'stainless-python-retry-a78b93fb-07c8-4ffd-86d5-f3cc2fbe6ced', 'json_data': {'model': 'rtlingo/mobiuslabsgmbh-faster-whisper-large-v3-turbo', 'response_format': 'json', 'stream': False}, 'extra_json': {'vad_filter': False}}
2025-09-24T15:44:09.446629673Z DEBUG:openai._base_client:Sending HTTP Request: POST http://speaches:8000/v1/audio/transcriptions
2025-09-24T15:44:09.447059377Z DEBUG:httpcore.connection:close.started
2025-09-24T15:44:09.447394165Z DEBUG:httpcore.connection:close.complete
2025-09-24T15:44:09.447545388Z DEBUG:httpcore.connection:connect_tcp.started host='speaches' port=8000 local_address=None timeout=5.0 socket_options=None
2025-09-24T15:44:09.449258395Z DEBUG:httpcore.connection:connect_tcp.complete return_value=<httpcore._backends.anyio.AnyIOStream object at 0x78efe719ee40>
2025-09-24T15:44:09.449268439Z DEBUG:httpcore.http11:send_request_headers.started request=<Request [b'POST']>
2025-09-24T15:44:09.449320445Z DEBUG:httpcore.http11:send_request_headers.complete
2025-09-24T15:44:09.449343169Z DEBUG:httpcore.http11:send_request_body.started request=<Request [b'POST']>
2025-09-24T15:44:09.450503180Z DEBUG:httpcore.http11:send_request_body.complete
2025-09-24T15:44:09.450557139Z DEBUG:httpcore.http11:receive_response_headers.started request=<Request [b'POST']>
2025-09-24T15:44:09.955307181Z DEBUG:httpcore.http11:receive_response_headers.complete return_value=(b'HTTP/1.1', 200, b'OK', [(b'date', b'Wed, 24 Sep 2025 15:44:08 GMT'), (b'server', b'uvicorn'), (b'content-length', b'39'), (b'content-type', b'application/json')])
2025-09-24T15:44:09.955358682Z INFO:httpx:HTTP Request: POST http://speaches:8000/v1/audio/transcriptions "HTTP/1.1 200 OK"
2025-09-24T15:44:09.955366829Z DEBUG:httpcore.http11:receive_response_body.started request=<Request [b'POST']>
2025-09-24T15:44:09.955474984Z DEBUG:httpcore.http11:receive_response_body.complete
2025-09-24T15:44:09.955484983Z DEBUG:httpcore.http11:response_closed.started
2025-09-24T15:44:09.955486650Z DEBUG:httpcore.http11:response_closed.complete
2025-09-24T15:44:09.955527681Z DEBUG:openai._base_client:HTTP Response: POST http://speaches:8000/v1/audio/transcriptions "200 OK" Headers({'date': 'Wed, 24 Sep 2025 15:44:08 GMT', 'server': 'uvicorn', 'content-length': '39', 'content-type': 'application/json'})
2025-09-24T15:44:09.955547715Z DEBUG:openai._base_client:request_id: None
2025-09-24T15:44:09.955716655Z DEBUG:wyoming_openai.handler:Outgoing event type transcript-start
2025-09-24T15:44:09.955822444Z DEBUG:wyoming_openai.handler:Handling non-streaming transcription response
2025-09-24T15:44:09.955829081Z INFO:wyoming_openai.handler:Successfully transcribed: Tell me another short story.
2025-09-24T15:44:09.955830338Z DEBUG:wyoming_openai.handler:Outgoing event type transcript
2025-09-24T15:44:09.955848881Z DEBUG:wyoming_openai.handler:Outgoing event type transcript-stop
2025-09-24T15:44:15.324833388Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-start
2025-09-24T15:44:15.324861658Z DEBUG:wyoming_openai.handler:Handling synthesize-start event: SynthesizeStart(voice=SynthesizeVoice(name='af_aoede', language=None, speaker=None), context=None)
2025-09-24T15:44:15.324885129Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:15.324903935Z DEBUG:wyoming_openai.handler:Received synthesis chunk: 'In a quiet hallway, a single light flickered once,...' (length: 61)
2025-09-24T15:44:15.324910601Z DEBUG:wyoming_openai.handler:Creating new pysbd segmenter for language: en
2025-09-24T15:44:15.325293022Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'In a quiet hallway, a single light flickered once, twice—then'
2025-09-24T15:44:15.376346085Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:15.376390514Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' stayed' (length: 7)
2025-09-24T15:44:15.376888760Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'In a quiet hallway, a single light flickered once, twice—then stayed'
2025-09-24T15:44:15.428540615Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:15.428570618Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' on' (length: 3)
2025-09-24T15:44:15.428946670Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'In a quiet hallway, a single light flickered once, twice—then stayed on'
2025-09-24T15:44:15.480573676Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:15.480610215Z DEBUG:wyoming_openai.handler:Received synthesis chunk: '.
2025-09-24T15:44:15.480614557Z 
2025-09-24T15:44:15.480617140Z ' (length: 3)
2025-09-24T15:44:15.481024644Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'In a quiet hallway, a single light flickered once, twice—then stayed on.
2025-09-24T15:44:15.481034881Z 
2025-09-24T15:44:15.481036530Z '
2025-09-24T15:44:15.533246070Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:15.533265869Z DEBUG:wyoming_openai.handler:Received synthesis chunk: 'It' (length: 2)
2025-09-24T15:44:15.533609985Z INFO:wyoming_openai.handler:Detected 1 ready sentences for immediate synthesis: ['In a quiet hallway, a single l...']
2025-09-24T15:44:15.533645898Z DEBUG:wyoming_openai.handler:Starting incremental synthesis for sentence 1: In a quiet hallway, a single light flickered once,...
2025-09-24T15:44:15.533859907Z DEBUG:openai._base_client:Request options: {'method': 'post', 'url': '/audio/speech', 'headers': {'Accept': 'application/octet-stream', 'X-Stainless-Raw-Response': 'stream'}, 'files': None, 'idempotency_key': 'stainless-python-retry-4282016c-925f-40da-a729-57c092c0fea5', 'json_data': {'input': 'In a quiet hallway, a single light flickered once, twice—then stayed on.', 'model': 'kokoro', 'voice': 'af_aoede'}}
2025-09-24T15:44:15.534113764Z DEBUG:openai._base_client:Sending HTTP Request: POST http://kokoro:8880/v1/audio/speech
2025-09-24T15:44:15.534272725Z DEBUG:httpcore.connection:close.started
2025-09-24T15:44:15.534371116Z DEBUG:httpcore.connection:close.complete
2025-09-24T15:44:15.534427152Z DEBUG:httpcore.connection:connect_tcp.started host='kokoro' port=8880 local_address=None timeout=5.0 socket_options=None
2025-09-24T15:44:15.535053475Z DEBUG:httpcore.connection:connect_tcp.complete return_value=<httpcore._backends.anyio.AnyIOStream object at 0x78efe719ea20>
2025-09-24T15:44:15.535132314Z DEBUG:httpcore.http11:send_request_headers.started request=<Request [b'POST']>
2025-09-24T15:44:15.535283008Z DEBUG:httpcore.http11:send_request_headers.complete
2025-09-24T15:44:15.535295217Z DEBUG:httpcore.http11:send_request_body.started request=<Request [b'POST']>
2025-09-24T15:44:15.535397190Z DEBUG:httpcore.http11:send_request_body.complete
2025-09-24T15:44:15.535423395Z DEBUG:httpcore.http11:receive_response_headers.started request=<Request [b'POST']>
2025-09-24T15:44:15.537471709Z DEBUG:httpcore.http11:receive_response_headers.complete return_value=(b'HTTP/1.1', 200, b'OK', [(b'date', b'Wed, 24 Sep 2025 15:44:14 GMT'), (b'server', b'uvicorn'), (b'content-disposition', b'attachment; filename=speech.mp3'), (b'x-accel-buffering', b'no'), (b'cache-control', b'no-cache'), (b'content-type', b'audio/mpeg'), (b'Transfer-Encoding', b'chunked')])
2025-09-24T15:44:15.537656087Z INFO:httpx:HTTP Request: POST http://kokoro:8880/v1/audio/speech "HTTP/1.1 200 OK"
2025-09-24T15:44:15.537714865Z DEBUG:openai._base_client:HTTP Response: POST http://kokoro:8880/v1/audio/speech "200 OK" Headers({'date': 'Wed, 24 Sep 2025 15:44:14 GMT', 'server': 'uvicorn', 'content-disposition': 'attachment; filename=speech.mp3', 'x-accel-buffering': 'no', 'cache-control': 'no-cache', 'content-type': 'audio/mpeg', 'transfer-encoding': 'chunked'})
2025-09-24T15:44:15.537758652Z DEBUG:openai._base_client:request_id: None
2025-09-24T15:44:15.537843891Z DEBUG:httpcore.http11:receive_response_body.started request=<Request [b'POST']>
2025-09-24T15:44:15.662258518Z DEBUG:httpcore.http11:receive_response_body.complete
2025-09-24T15:44:15.662279023Z DEBUG:httpcore.http11:response_closed.started
2025-09-24T15:44:15.662287613Z DEBUG:httpcore.http11:response_closed.complete
2025-09-24T15:44:15.662389140Z DEBUG:wyoming_openai.handler:Completed synthesis for chunk: In a quiet hallway, a single light flickered once,...
2025-09-24T15:44:15.662399872Z DEBUG:wyoming_openai.handler:Failed to parse WAV header: file does not start with RIFF id
2025-09-24T15:44:15.662402894Z DEBUG:wyoming_openai.handler:Could not parse WAV header, using defaults: 24000 Hz
2025-09-24T15:44:15.662407218Z DEBUG:wyoming_openai.handler:Outgoing event type audio-start
2025-09-24T15:44:15.662497358Z DEBUG:wyoming_openai.handler:Outgoing event type audio-chunk
2025-09-24T15:44:15.662536767Z DEBUG:wyoming_openai.handler:Outgoing event type audio-chunk (subsequent audio chunks will not be logged)
2025-09-24T15:44:15.662976327Z DEBUG:wyoming_openai.handler:Successfully streamed incremental sentence 1, timestamp: 1432.92
2025-09-24T15:44:15.663007893Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:15.663036829Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' wasn' (length: 5)
2025-09-24T15:44:15.663197282Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'It wasn'
2025-09-24T15:44:15.663264804Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:15.663274691Z DEBUG:wyoming_openai.handler:Received synthesis chunk: '’t' (length: 2)
2025-09-24T15:44:15.663395058Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'It wasn’t'
2025-09-24T15:44:15.720722945Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:15.720743170Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' broken' (length: 7)
2025-09-24T15:44:15.721051311Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'It wasn’t broken'
2025-09-24T15:44:15.773102989Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:15.773123061Z DEBUG:wyoming_openai.handler:Received synthesis chunk: '.' (length: 1)
2025-09-24T15:44:15.773416934Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'It wasn’t broken.'
2025-09-24T15:44:15.826841807Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:15.826864790Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' It' (length: 3)
2025-09-24T15:44:15.827147321Z INFO:wyoming_openai.handler:Detected 1 ready sentences for immediate synthesis: ['It wasn’t broken.']
2025-09-24T15:44:15.827156386Z DEBUG:wyoming_openai.handler:Starting incremental synthesis for sentence 1: It wasn’t broken.
2025-09-24T15:44:15.827398104Z DEBUG:openai._base_client:Request options: {'method': 'post', 'url': '/audio/speech', 'headers': {'Accept': 'application/octet-stream', 'X-Stainless-Raw-Response': 'stream'}, 'files': None, 'idempotency_key': 'stainless-python-retry-742564a6-fbfc-416d-ad8f-d980f51624a3', 'json_data': {'input': 'It wasn’t broken.', 'model': 'kokoro', 'voice': 'af_aoede'}}
2025-09-24T15:44:15.827606913Z DEBUG:openai._base_client:Sending HTTP Request: POST http://kokoro:8880/v1/audio/speech
2025-09-24T15:44:15.827834741Z DEBUG:httpcore.http11:send_request_headers.started request=<Request [b'POST']>
2025-09-24T15:44:15.827998467Z DEBUG:httpcore.http11:send_request_headers.complete
2025-09-24T15:44:15.828025654Z DEBUG:httpcore.http11:send_request_body.started request=<Request [b'POST']>
2025-09-24T15:44:15.828079871Z DEBUG:httpcore.http11:send_request_body.complete
2025-09-24T15:44:15.828091083Z DEBUG:httpcore.http11:receive_response_headers.started request=<Request [b'POST']>
2025-09-24T15:44:15.830086346Z DEBUG:httpcore.http11:receive_response_headers.complete return_value=(b'HTTP/1.1', 200, b'OK', [(b'date', b'Wed, 24 Sep 2025 15:44:14 GMT'), (b'server', b'uvicorn'), (b'content-disposition', b'attachment; filename=speech.mp3'), (b'x-accel-buffering', b'no'), (b'cache-control', b'no-cache'), (b'content-type', b'audio/mpeg'), (b'Transfer-Encoding', b'chunked')])
2025-09-24T15:44:15.830256793Z INFO:httpx:HTTP Request: POST http://kokoro:8880/v1/audio/speech "HTTP/1.1 200 OK"
2025-09-24T15:44:15.830431423Z DEBUG:openai._base_client:HTTP Response: POST http://kokoro:8880/v1/audio/speech "200 OK" Headers({'date': 'Wed, 24 Sep 2025 15:44:14 GMT', 'server': 'uvicorn', 'content-disposition': 'attachment; filename=speech.mp3', 'x-accel-buffering': 'no', 'cache-control': 'no-cache', 'content-type': 'audio/mpeg', 'transfer-encoding': 'chunked'})
2025-09-24T15:44:15.830447066Z DEBUG:openai._base_client:request_id: None
2025-09-24T15:44:15.830450940Z DEBUG:httpcore.http11:receive_response_body.started request=<Request [b'POST']>
2025-09-24T15:44:15.920097824Z DEBUG:httpcore.http11:receive_response_body.complete
2025-09-24T15:44:15.920127737Z DEBUG:httpcore.http11:response_closed.started
2025-09-24T15:44:15.920172317Z DEBUG:httpcore.http11:response_closed.complete
2025-09-24T15:44:15.920256774Z DEBUG:wyoming_openai.handler:Completed synthesis for chunk: It wasn’t broken.
2025-09-24T15:44:15.920289735Z DEBUG:wyoming_openai.handler:Failed to parse WAV header: file does not start with RIFF id
2025-09-24T15:44:15.920296531Z DEBUG:wyoming_openai.handler:Could not parse WAV header, using defaults: 24000 Hz
2025-09-24T15:44:15.920667038Z DEBUG:wyoming_openai.handler:Successfully streamed incremental sentence 1, timestamp: 1905.83
2025-09-24T15:44:15.920694116Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:15.920714527Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' was' (length: 4)
2025-09-24T15:44:15.920952116Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'It was'
2025-09-24T15:44:15.952184948Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:15.952202738Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' watching' (length: 9)
2025-09-24T15:44:15.952584536Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'It was watching'
2025-09-24T15:44:16.004597366Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:16.004627291Z DEBUG:wyoming_openai.handler:Received synthesis chunk: '.
2025-09-24T15:44:16.004630124Z 
2025-09-24T15:44:16.004631754Z ' (length: 3)
2025-09-24T15:44:16.004973831Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'It was watching.
2025-09-24T15:44:16.004987159Z 
2025-09-24T15:44:16.004989622Z '
2025-09-24T15:44:16.060161196Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:16.060187647Z DEBUG:wyoming_openai.handler:Received synthesis chunk: 'Every' (length: 5)
2025-09-24T15:44:16.060552123Z INFO:wyoming_openai.handler:Detected 1 ready sentences for immediate synthesis: ['It was watching.']
2025-09-24T15:44:16.060578360Z DEBUG:wyoming_openai.handler:Starting incremental synthesis for sentence 1: It was watching.
2025-09-24T15:44:16.060884770Z DEBUG:openai._base_client:Request options: {'method': 'post', 'url': '/audio/speech', 'headers': {'Accept': 'application/octet-stream', 'X-Stainless-Raw-Response': 'stream'}, 'files': None, 'idempotency_key': 'stainless-python-retry-e35adfb8-4a30-44e9-b1ad-03f94307d5f1', 'json_data': {'input': 'It was watching.', 'model': 'kokoro', 'voice': 'af_aoede'}}
2025-09-24T15:44:16.061106209Z DEBUG:openai._base_client:Sending HTTP Request: POST http://kokoro:8880/v1/audio/speech
2025-09-24T15:44:16.061312541Z DEBUG:httpcore.http11:send_request_headers.started request=<Request [b'POST']>
2025-09-24T15:44:16.061472984Z DEBUG:httpcore.http11:send_request_headers.complete
2025-09-24T15:44:16.061496307Z DEBUG:httpcore.http11:send_request_body.started request=<Request [b'POST']>
2025-09-24T15:44:16.061587062Z DEBUG:httpcore.http11:send_request_body.complete
2025-09-24T15:44:16.061602089Z DEBUG:httpcore.http11:receive_response_headers.started request=<Request [b'POST']>
2025-09-24T15:44:16.063197665Z DEBUG:httpcore.http11:receive_response_headers.complete return_value=(b'HTTP/1.1', 200, b'OK', [(b'date', b'Wed, 24 Sep 2025 15:44:15 GMT'), (b'server', b'uvicorn'), (b'content-disposition', b'attachment; filename=speech.mp3'), (b'x-accel-buffering', b'no'), (b'cache-control', b'no-cache'), (b'content-type', b'audio/mpeg'), (b'Transfer-Encoding', b'chunked')])
2025-09-24T15:44:16.063339080Z INFO:httpx:HTTP Request: POST http://kokoro:8880/v1/audio/speech "HTTP/1.1 200 OK"
2025-09-24T15:44:16.063390520Z DEBUG:openai._base_client:HTTP Response: POST http://kokoro:8880/v1/audio/speech "200 OK" Headers({'date': 'Wed, 24 Sep 2025 15:44:15 GMT', 'server': 'uvicorn', 'content-disposition': 'attachment; filename=speech.mp3', 'x-accel-buffering': 'no', 'cache-control': 'no-cache', 'content-type': 'audio/mpeg', 'transfer-encoding': 'chunked'})
2025-09-24T15:44:16.063409374Z DEBUG:openai._base_client:request_id: None
2025-09-24T15:44:16.063495516Z DEBUG:httpcore.http11:receive_response_body.started request=<Request [b'POST']>
2025-09-24T15:44:16.146709121Z DEBUG:httpcore.http11:receive_response_body.complete
2025-09-24T15:44:16.146741733Z DEBUG:httpcore.http11:response_closed.started
2025-09-24T15:44:16.146812593Z DEBUG:httpcore.http11:response_closed.complete
2025-09-24T15:44:16.146872120Z DEBUG:wyoming_openai.handler:Completed synthesis for chunk: It was watching.
2025-09-24T15:44:16.146901519Z DEBUG:wyoming_openai.handler:Failed to parse WAV header: file does not start with RIFF id
2025-09-24T15:44:16.146919840Z DEBUG:wyoming_openai.handler:Could not parse WAV header, using defaults: 24000 Hz
2025-09-24T15:44:16.147255022Z DEBUG:wyoming_openai.handler:Successfully streamed incremental sentence 1, timestamp: 2338.75
2025-09-24T15:44:16.147290627Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:16.147308563Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' night' (length: 6)
2025-09-24T15:44:16.147598125Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'Every night'
2025-09-24T15:44:16.179293344Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:16.179325478Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' at' (length: 3)
2025-09-24T15:44:16.179601430Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'Every night at'
2025-09-24T15:44:16.230570438Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:16.230590699Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' ' (length: 1)
2025-09-24T15:44:16.230828275Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'Every night at '
2025-09-24T15:44:16.282359239Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:16.282385607Z DEBUG:wyoming_openai.handler:Received synthesis chunk: '1' (length: 1)
2025-09-24T15:44:16.282707292Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'Every night at 1'
2025-09-24T15:44:16.334847015Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:16.334876484Z DEBUG:wyoming_openai.handler:Received synthesis chunk: '0' (length: 1)
2025-09-24T15:44:16.335095202Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'Every night at 10'
2025-09-24T15:44:16.386048137Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:16.386070090Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ':' (length: 1)
2025-09-24T15:44:16.386441187Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'Every night at 10:'
2025-09-24T15:44:16.438877132Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:16.438905436Z DEBUG:wyoming_openai.handler:Received synthesis chunk: '4' (length: 1)
2025-09-24T15:44:16.439339840Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'Every night at 10:4'
2025-09-24T15:44:16.490646262Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:16.490677990Z DEBUG:wyoming_openai.handler:Received synthesis chunk: '7' (length: 1)
2025-09-24T15:44:16.491027281Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'Every night at 10:47'
2025-09-24T15:44:16.543483876Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:16.543505754Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ',' (length: 1)
2025-09-24T15:44:16.543853061Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'Every night at 10:47,'
2025-09-24T15:44:16.596164590Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:16.596182557Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' the' (length: 4)
2025-09-24T15:44:16.596493586Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'Every night at 10:47, the'
2025-09-24T15:44:16.648805548Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:16.648831736Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' shadow' (length: 7)
2025-09-24T15:44:16.649103943Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'Every night at 10:47, the shadow'
2025-09-24T15:44:16.701196311Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:16.701225318Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' of' (length: 3)
2025-09-24T15:44:16.701607655Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'Every night at 10:47, the shadow of'
2025-09-24T15:44:16.753930252Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:16.753949026Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' a' (length: 2)
2025-09-24T15:44:16.754268710Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'Every night at 10:47, the shadow of a'
2025-09-24T15:44:16.807170670Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:16.807194632Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' child' (length: 6)
2025-09-24T15:44:16.807587283Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'Every night at 10:47, the shadow of a child'
2025-09-24T15:44:16.860260492Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:16.860278385Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' would' (length: 6)
2025-09-24T15:44:16.860747747Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'Every night at 10:47, the shadow of a child would'
2025-09-24T15:44:16.913176665Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:16.913206586Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' appear' (length: 7)
2025-09-24T15:44:16.913514301Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'Every night at 10:47, the shadow of a child would appear'
2025-09-24T15:44:16.965232892Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:16.965250943Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' on' (length: 3)
2025-09-24T15:44:16.965592705Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'Every night at 10:47, the shadow of a child would appear on'
2025-09-24T15:44:17.021067253Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:17.021085618Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' the' (length: 4)
2025-09-24T15:44:17.021558601Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'Every night at 10:47, the shadow of a child would appear on the'
2025-09-24T15:44:17.073612454Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:17.073632098Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' wall' (length: 5)
2025-09-24T15:44:17.073956486Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'Every night at 10:47, the shadow of a child would appear on the wall'
2025-09-24T15:44:17.126947833Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:17.126972411Z DEBUG:wyoming_openai.handler:Received synthesis chunk: '.' (length: 1)
2025-09-24T15:44:17.127465663Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'Every night at 10:47, the shadow of a child would appear on the wall.'
2025-09-24T15:44:17.179381490Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:17.179400404Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' But' (length: 4)
2025-09-24T15:44:17.179719489Z INFO:wyoming_openai.handler:Detected 1 ready sentences for immediate synthesis: ['Every night at 10:47, the shad...']
2025-09-24T15:44:17.179740653Z DEBUG:wyoming_openai.handler:Starting incremental synthesis for sentence 1: Every night at 10:47, the shadow of a child would ...
2025-09-24T15:44:17.179968626Z DEBUG:openai._base_client:Request options: {'method': 'post', 'url': '/audio/speech', 'headers': {'Accept': 'application/octet-stream', 'X-Stainless-Raw-Response': 'stream'}, 'files': None, 'idempotency_key': 'stainless-python-retry-51e8b6c9-90d9-4c35-9641-4b63bedcf1fb', 'json_data': {'input': 'Every night at 10:47, the shadow of a child would appear on the wall.', 'model': 'kokoro', 'voice': 'af_aoede'}}
2025-09-24T15:44:17.180186690Z DEBUG:openai._base_client:Sending HTTP Request: POST http://kokoro:8880/v1/audio/speech
2025-09-24T15:44:17.180382667Z DEBUG:httpcore.http11:send_request_headers.started request=<Request [b'POST']>
2025-09-24T15:44:17.180576283Z DEBUG:httpcore.http11:send_request_headers.complete
2025-09-24T15:44:17.180608141Z DEBUG:httpcore.http11:send_request_body.started request=<Request [b'POST']>
2025-09-24T15:44:17.180726747Z DEBUG:httpcore.http11:send_request_body.complete
2025-09-24T15:44:17.180737071Z DEBUG:httpcore.http11:receive_response_headers.started request=<Request [b'POST']>
2025-09-24T15:44:17.182238379Z DEBUG:httpcore.http11:receive_response_headers.complete return_value=(b'HTTP/1.1', 200, b'OK', [(b'date', b'Wed, 24 Sep 2025 15:44:16 GMT'), (b'server', b'uvicorn'), (b'content-disposition', b'attachment; filename=speech.mp3'), (b'x-accel-buffering', b'no'), (b'cache-control', b'no-cache'), (b'content-type', b'audio/mpeg'), (b'Transfer-Encoding', b'chunked')])
2025-09-24T15:44:17.182388202Z INFO:httpx:HTTP Request: POST http://kokoro:8880/v1/audio/speech "HTTP/1.1 200 OK"
2025-09-24T15:44:17.182424756Z DEBUG:openai._base_client:HTTP Response: POST http://kokoro:8880/v1/audio/speech "200 OK" Headers({'date': 'Wed, 24 Sep 2025 15:44:16 GMT', 'server': 'uvicorn', 'content-disposition': 'attachment; filename=speech.mp3', 'x-accel-buffering': 'no', 'cache-control': 'no-cache', 'content-type': 'audio/mpeg', 'transfer-encoding': 'chunked'})
2025-09-24T15:44:17.182463891Z DEBUG:openai._base_client:request_id: None
2025-09-24T15:44:17.182522888Z DEBUG:httpcore.http11:receive_response_body.started request=<Request [b'POST']>
2025-09-24T15:44:17.300487618Z DEBUG:httpcore.http11:receive_response_body.complete
2025-09-24T15:44:17.300523523Z DEBUG:httpcore.http11:response_closed.started
2025-09-24T15:44:17.300562550Z DEBUG:httpcore.http11:response_closed.complete
2025-09-24T15:44:17.300650752Z DEBUG:wyoming_openai.handler:Completed synthesis for chunk: Every night at 10:47, the shadow of a child would ...
2025-09-24T15:44:17.300680119Z DEBUG:wyoming_openai.handler:Failed to parse WAV header: file does not start with RIFF id
2025-09-24T15:44:17.300697842Z DEBUG:wyoming_openai.handler:Could not parse WAV header, using defaults: 24000 Hz
2025-09-24T15:44:17.301228345Z DEBUG:wyoming_openai.handler:Successfully streamed incremental sentence 1, timestamp: 3827.67
2025-09-24T15:44:17.301336338Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:17.301339319Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' no' (length: 3)
2025-09-24T15:44:17.301528599Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'But no'
2025-09-24T15:44:17.313806678Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:17.313831046Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' child' (length: 6)
2025-09-24T15:44:17.314056620Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'But no child'
2025-09-24T15:44:17.368995298Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:17.369016740Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' lived' (length: 6)
2025-09-24T15:44:17.369271856Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'But no child lived'
2025-09-24T15:44:17.421270319Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:17.421299461Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' there' (length: 6)
2025-09-24T15:44:17.421690081Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'But no child lived there'
2025-09-24T15:44:17.473317409Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:17.473341381Z DEBUG:wyoming_openai.handler:Received synthesis chunk: '.
2025-09-24T15:44:17.473344124Z 
2025-09-24T15:44:17.473345810Z ' (length: 3)
2025-09-24T15:44:17.473765557Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'But no child lived there.
2025-09-24T15:44:17.473786796Z 
2025-09-24T15:44:17.473788870Z '
2025-09-24T15:44:17.524906926Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:17.524928379Z DEBUG:wyoming_openai.handler:Received synthesis chunk: 'One' (length: 3)
2025-09-24T15:44:17.525194064Z INFO:wyoming_openai.handler:Detected 1 ready sentences for immediate synthesis: ['But no child lived there.']
2025-09-24T15:44:17.525201897Z DEBUG:wyoming_openai.handler:Starting incremental synthesis for sentence 1: But no child lived there.
2025-09-24T15:44:17.525441363Z DEBUG:openai._base_client:Request options: {'method': 'post', 'url': '/audio/speech', 'headers': {'Accept': 'application/octet-stream', 'X-Stainless-Raw-Response': 'stream'}, 'files': None, 'idempotency_key': 'stainless-python-retry-125525d3-3c3f-465c-972d-2f37e2489b27', 'json_data': {'input': 'But no child lived there.', 'model': 'kokoro', 'voice': 'af_aoede'}}
2025-09-24T15:44:17.525651014Z DEBUG:openai._base_client:Sending HTTP Request: POST http://kokoro:8880/v1/audio/speech
2025-09-24T15:44:17.525853574Z DEBUG:httpcore.http11:send_request_headers.started request=<Request [b'POST']>
2025-09-24T15:44:17.525998592Z DEBUG:httpcore.http11:send_request_headers.complete
2025-09-24T15:44:17.526038656Z DEBUG:httpcore.http11:send_request_body.started request=<Request [b'POST']>
2025-09-24T15:44:17.526092127Z DEBUG:httpcore.http11:send_request_body.complete
2025-09-24T15:44:17.526100656Z DEBUG:httpcore.http11:receive_response_headers.started request=<Request [b'POST']>
2025-09-24T15:44:17.527875000Z DEBUG:httpcore.http11:receive_response_headers.complete return_value=(b'HTTP/1.1', 200, b'OK', [(b'date', b'Wed, 24 Sep 2025 15:44:16 GMT'), (b'server', b'uvicorn'), (b'content-disposition', b'attachment; filename=speech.mp3'), (b'x-accel-buffering', b'no'), (b'cache-control', b'no-cache'), (b'content-type', b'audio/mpeg'), (b'Transfer-Encoding', b'chunked')])
2025-09-24T15:44:17.528057554Z INFO:httpx:HTTP Request: POST http://kokoro:8880/v1/audio/speech "HTTP/1.1 200 OK"
2025-09-24T15:44:17.528108999Z DEBUG:openai._base_client:HTTP Response: POST http://kokoro:8880/v1/audio/speech "200 OK" Headers({'date': 'Wed, 24 Sep 2025 15:44:16 GMT', 'server': 'uvicorn', 'content-disposition': 'attachment; filename=speech.mp3', 'x-accel-buffering': 'no', 'cache-control': 'no-cache', 'content-type': 'audio/mpeg', 'transfer-encoding': 'chunked'})
2025-09-24T15:44:17.528134947Z DEBUG:openai._base_client:request_id: None
2025-09-24T15:44:17.528201087Z DEBUG:httpcore.http11:receive_response_body.started request=<Request [b'POST']>
2025-09-24T15:44:17.593643679Z DEBUG:httpcore.http11:receive_response_body.complete
2025-09-24T15:44:17.593680403Z DEBUG:httpcore.http11:response_closed.started
2025-09-24T15:44:17.593717966Z DEBUG:httpcore.http11:response_closed.complete
2025-09-24T15:44:17.593794988Z DEBUG:wyoming_openai.handler:Completed synthesis for chunk: But no child lived there.
2025-09-24T15:44:17.593820626Z DEBUG:wyoming_openai.handler:Failed to parse WAV header: file does not start with RIFF id
2025-09-24T15:44:17.593830712Z DEBUG:wyoming_openai.handler:Could not parse WAV header, using defaults: 24000 Hz
2025-09-24T15:44:17.594264698Z DEBUG:wyoming_openai.handler:Successfully streamed incremental sentence 1, timestamp: 4388.58
2025-09-24T15:44:17.594322087Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:17.594336350Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' evening' (length: 8)
2025-09-24T15:44:17.594639538Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'One evening'
2025-09-24T15:44:17.647238956Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:17.647272494Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ',' (length: 1)
2025-09-24T15:44:17.647713880Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'One evening,'
2025-09-24T15:44:17.699664152Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:17.699693685Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' the' (length: 4)
2025-09-24T15:44:17.699900916Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'One evening, the'
2025-09-24T15:44:17.751049978Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:17.751082295Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' light' (length: 6)
2025-09-24T15:44:17.751344777Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'One evening, the light'
2025-09-24T15:44:17.803218987Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:17.803249739Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' blinked' (length: 8)
2025-09-24T15:44:17.803651736Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'One evening, the light blinked'
2025-09-24T15:44:17.856039555Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:17.856056001Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' slowly' (length: 7)
2025-09-24T15:44:17.856424566Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'One evening, the light blinked slowly'
2025-09-24T15:44:17.908462527Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:17.908486412Z DEBUG:wyoming_openai.handler:Received synthesis chunk: '—' (length: 1)
2025-09-24T15:44:17.908853365Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'One evening, the light blinked slowly—'
2025-09-24T15:44:17.962266739Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:17.962287163Z DEBUG:wyoming_openai.handler:Received synthesis chunk: 'once' (length: 4)
2025-09-24T15:44:17.962759291Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'One evening, the light blinked slowly—once'
2025-09-24T15:44:18.015251636Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:18.015269249Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ',' (length: 1)
2025-09-24T15:44:18.015720446Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'One evening, the light blinked slowly—once,'
2025-09-24T15:44:18.067704312Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:18.067731608Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' twice' (length: 6)
2025-09-24T15:44:18.068037258Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'One evening, the light blinked slowly—once, twice'
2025-09-24T15:44:18.121281975Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:18.121300950Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ',' (length: 1)
2025-09-24T15:44:18.121739968Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'One evening, the light blinked slowly—once, twice,'
2025-09-24T15:44:18.175348365Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:18.175371415Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' three' (length: 6)
2025-09-24T15:44:18.175868095Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'One evening, the light blinked slowly—once, twice, three'
2025-09-24T15:44:18.228398824Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:18.228418413Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' times' (length: 6)
2025-09-24T15:44:18.228769800Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'One evening, the light blinked slowly—once, twice, three times'
2025-09-24T15:44:18.281252751Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:18.281275715Z DEBUG:wyoming_openai.handler:Received synthesis chunk: '.
2025-09-24T15:44:18.281278269Z 
2025-09-24T15:44:18.281279896Z ' (length: 3)
2025-09-24T15:44:18.281616158Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'One evening, the light blinked slowly—once, twice, three times.
2025-09-24T15:44:18.281627291Z 
2025-09-24T15:44:18.281629184Z '
2025-09-24T15:44:18.333951523Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:18.333978572Z DEBUG:wyoming_openai.handler:Received synthesis chunk: 'And' (length: 3)
2025-09-24T15:44:18.334275613Z INFO:wyoming_openai.handler:Detected 1 ready sentences for immediate synthesis: ['One evening, the light blinked...']
2025-09-24T15:44:18.334298178Z DEBUG:wyoming_openai.handler:Starting incremental synthesis for sentence 1: One evening, the light blinked slowly—once, twice,...
2025-09-24T15:44:18.334521916Z DEBUG:openai._base_client:Request options: {'method': 'post', 'url': '/audio/speech', 'headers': {'Accept': 'application/octet-stream', 'X-Stainless-Raw-Response': 'stream'}, 'files': None, 'idempotency_key': 'stainless-python-retry-13a24127-fe88-4188-903d-ce0f148e1dbc', 'json_data': {'input': 'One evening, the light blinked slowly—once, twice, three times.', 'model': 'kokoro', 'voice': 'af_aoede'}}
2025-09-24T15:44:18.334757535Z DEBUG:openai._base_client:Sending HTTP Request: POST http://kokoro:8880/v1/audio/speech
2025-09-24T15:44:18.334953043Z DEBUG:httpcore.http11:send_request_headers.started request=<Request [b'POST']>
2025-09-24T15:44:18.335100115Z DEBUG:httpcore.http11:send_request_headers.complete
2025-09-24T15:44:18.335111302Z DEBUG:httpcore.http11:send_request_body.started request=<Request [b'POST']>
2025-09-24T15:44:18.335192597Z DEBUG:httpcore.http11:send_request_body.complete
2025-09-24T15:44:18.335217366Z DEBUG:httpcore.http11:receive_response_headers.started request=<Request [b'POST']>
2025-09-24T15:44:18.336889882Z DEBUG:httpcore.http11:receive_response_headers.complete return_value=(b'HTTP/1.1', 200, b'OK', [(b'date', b'Wed, 24 Sep 2025 15:44:18 GMT'), (b'server', b'uvicorn'), (b'content-disposition', b'attachment; filename=speech.mp3'), (b'x-accel-buffering', b'no'), (b'cache-control', b'no-cache'), (b'content-type', b'audio/mpeg'), (b'Transfer-Encoding', b'chunked')])
2025-09-24T15:44:18.337040725Z INFO:httpx:HTTP Request: POST http://kokoro:8880/v1/audio/speech "HTTP/1.1 200 OK"
2025-09-24T15:44:18.337071782Z DEBUG:openai._base_client:HTTP Response: POST http://kokoro:8880/v1/audio/speech "200 OK" Headers({'date': 'Wed, 24 Sep 2025 15:44:18 GMT', 'server': 'uvicorn', 'content-disposition': 'attachment; filename=speech.mp3', 'x-accel-buffering': 'no', 'cache-control': 'no-cache', 'content-type': 'audio/mpeg', 'transfer-encoding': 'chunked'})
2025-09-24T15:44:18.337092948Z DEBUG:openai._base_client:request_id: None
2025-09-24T15:44:18.337160630Z DEBUG:httpcore.http11:receive_response_body.started request=<Request [b'POST']>
2025-09-24T15:44:18.450043051Z DEBUG:httpcore.http11:receive_response_body.complete
2025-09-24T15:44:18.450058492Z DEBUG:httpcore.http11:response_closed.started
2025-09-24T15:44:18.450061010Z DEBUG:httpcore.http11:response_closed.complete
2025-09-24T15:44:18.450136816Z DEBUG:wyoming_openai.handler:Completed synthesis for chunk: One evening, the light blinked slowly—once, twice,...
2025-09-24T15:44:18.450150175Z DEBUG:wyoming_openai.handler:Failed to parse WAV header: file does not start with RIFF id
2025-09-24T15:44:18.450157406Z DEBUG:wyoming_openai.handler:Could not parse WAV header, using defaults: 24000 Hz
2025-09-24T15:44:18.450690955Z DEBUG:wyoming_openai.handler:Successfully streamed incremental sentence 1, timestamp: 5581.50
2025-09-24T15:44:18.450777801Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:18.450786390Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' the' (length: 4)
2025-09-24T15:44:18.451039408Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'And the'
2025-09-24T15:44:18.463471027Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:18.463494148Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' shadow' (length: 7)
2025-09-24T15:44:18.463776660Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'And the shadow'
2025-09-24T15:44:18.518202435Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:18.518225620Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' waved' (length: 6)
2025-09-24T15:44:18.518494325Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'And the shadow waved'
2025-09-24T15:44:18.570997298Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:18.571019734Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' back' (length: 5)
2025-09-24T15:44:18.571357789Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'And the shadow waved back'
2025-09-24T15:44:18.623861527Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:18.623885933Z DEBUG:wyoming_openai.handler:Received synthesis chunk: '.
2025-09-24T15:44:18.623888604Z 
2025-09-24T15:44:18.623890241Z ' (length: 3)
2025-09-24T15:44:18.624123964Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'And the shadow waved back.
2025-09-24T15:44:18.624133857Z 
2025-09-24T15:44:18.624135180Z '
2025-09-24T15:44:18.677954525Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:18.677977803Z DEBUG:wyoming_openai.handler:Received synthesis chunk: 'From' (length: 4)
2025-09-24T15:44:18.678309974Z INFO:wyoming_openai.handler:Detected 1 ready sentences for immediate synthesis: ['And the shadow waved back.']
2025-09-24T15:44:18.678317326Z DEBUG:wyoming_openai.handler:Starting incremental synthesis for sentence 1: And the shadow waved back.
2025-09-24T15:44:18.678561565Z DEBUG:openai._base_client:Request options: {'method': 'post', 'url': '/audio/speech', 'headers': {'Accept': 'application/octet-stream', 'X-Stainless-Raw-Response': 'stream'}, 'files': None, 'idempotency_key': 'stainless-python-retry-c9807edc-e970-4364-8716-3d8e38e4db6d', 'json_data': {'input': 'And the shadow waved back.', 'model': 'kokoro', 'voice': 'af_aoede'}}
2025-09-24T15:44:18.678808095Z DEBUG:openai._base_client:Sending HTTP Request: POST http://kokoro:8880/v1/audio/speech
2025-09-24T15:44:18.679028231Z DEBUG:httpcore.http11:send_request_headers.started request=<Request [b'POST']>
2025-09-24T15:44:18.679203942Z DEBUG:httpcore.http11:send_request_headers.complete
2025-09-24T15:44:18.679217156Z DEBUG:httpcore.http11:send_request_body.started request=<Request [b'POST']>
2025-09-24T15:44:18.679306580Z DEBUG:httpcore.http11:send_request_body.complete
2025-09-24T15:44:18.679334707Z DEBUG:httpcore.http11:receive_response_headers.started request=<Request [b'POST']>
2025-09-24T15:44:18.681015214Z DEBUG:httpcore.http11:receive_response_headers.complete return_value=(b'HTTP/1.1', 200, b'OK', [(b'date', b'Wed, 24 Sep 2025 15:44:18 GMT'), (b'server', b'uvicorn'), (b'content-disposition', b'attachment; filename=speech.mp3'), (b'x-accel-buffering', b'no'), (b'cache-control', b'no-cache'), (b'content-type', b'audio/mpeg'), (b'Transfer-Encoding', b'chunked')])
2025-09-24T15:44:18.681162058Z INFO:httpx:HTTP Request: POST http://kokoro:8880/v1/audio/speech "HTTP/1.1 200 OK"
2025-09-24T15:44:18.681198908Z DEBUG:openai._base_client:HTTP Response: POST http://kokoro:8880/v1/audio/speech "200 OK" Headers({'date': 'Wed, 24 Sep 2025 15:44:18 GMT', 'server': 'uvicorn', 'content-disposition': 'attachment; filename=speech.mp3', 'x-accel-buffering': 'no', 'cache-control': 'no-cache', 'content-type': 'audio/mpeg', 'transfer-encoding': 'chunked'})
2025-09-24T15:44:18.681214755Z DEBUG:openai._base_client:request_id: None
2025-09-24T15:44:18.681285333Z DEBUG:httpcore.http11:receive_response_body.started request=<Request [b'POST']>
2025-09-24T15:44:18.763833783Z DEBUG:httpcore.http11:receive_response_body.complete
2025-09-24T15:44:18.763925529Z DEBUG:httpcore.http11:response_closed.started
2025-09-24T15:44:18.763955176Z DEBUG:httpcore.http11:response_closed.complete
2025-09-24T15:44:18.764127025Z DEBUG:wyoming_openai.handler:Completed synthesis for chunk: And the shadow waved back.
2025-09-24T15:44:18.764135718Z DEBUG:wyoming_openai.handler:Failed to parse WAV header: file does not start with RIFF id
2025-09-24T15:44:18.764138019Z DEBUG:wyoming_openai.handler:Could not parse WAV header, using defaults: 24000 Hz
2025-09-24T15:44:18.764931708Z DEBUG:wyoming_openai.handler:Successfully streamed incremental sentence 1, timestamp: 6142.42
2025-09-24T15:44:18.764988479Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:18.765010559Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' then' (length: 5)
2025-09-24T15:44:18.765361538Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'From then'
2025-09-24T15:44:18.796083585Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:18.796108353Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' on' (length: 3)
2025-09-24T15:44:18.796466926Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'From then on'
2025-09-24T15:44:18.849124962Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:18.849149067Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ',' (length: 1)
2025-09-24T15:44:18.849552997Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'From then on,'
2025-09-24T15:44:18.902837303Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:18.902858996Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' the' (length: 4)
2025-09-24T15:44:18.903218264Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'From then on, the'
2025-09-24T15:44:18.956258025Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:18.956279996Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' light' (length: 6)
2025-09-24T15:44:18.956602929Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'From then on, the light'
2025-09-24T15:44:19.009917146Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:19.009947040Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' never' (length: 6)
2025-09-24T15:44:19.010621851Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'From then on, the light never'
2025-09-24T15:44:19.062813499Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:19.062834228Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' turned' (length: 7)
2025-09-24T15:44:19.063116639Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'From then on, the light never turned'
2025-09-24T15:44:19.114840648Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:19.114860937Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' off' (length: 4)
2025-09-24T15:44:19.115174297Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'From then on, the light never turned off'
2025-09-24T15:44:19.167405385Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:19.167425716Z DEBUG:wyoming_openai.handler:Received synthesis chunk: '.' (length: 1)
2025-09-24T15:44:19.167968633Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'From then on, the light never turned off.'
2025-09-24T15:44:19.219886019Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:19.219909712Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' And' (length: 4)
2025-09-24T15:44:19.220228933Z INFO:wyoming_openai.handler:Detected 1 ready sentences for immediate synthesis: ['From then on, the light never ...']
2025-09-24T15:44:19.220240610Z DEBUG:wyoming_openai.handler:Starting incremental synthesis for sentence 1: From then on, the light never turned off.
2025-09-24T15:44:19.220487336Z DEBUG:openai._base_client:Request options: {'method': 'post', 'url': '/audio/speech', 'headers': {'Accept': 'application/octet-stream', 'X-Stainless-Raw-Response': 'stream'}, 'files': None, 'idempotency_key': 'stainless-python-retry-6cc23f84-ce4c-4d7b-a28b-2ec57095d507', 'json_data': {'input': 'From then on, the light never turned off.', 'model': 'kokoro', 'voice': 'af_aoede'}}
2025-09-24T15:44:19.220732593Z DEBUG:openai._base_client:Sending HTTP Request: POST http://kokoro:8880/v1/audio/speech
2025-09-24T15:44:19.220944138Z DEBUG:httpcore.http11:send_request_headers.started request=<Request [b'POST']>
2025-09-24T15:44:19.221141415Z DEBUG:httpcore.http11:send_request_headers.complete
2025-09-24T15:44:19.221151343Z DEBUG:httpcore.http11:send_request_body.started request=<Request [b'POST']>
2025-09-24T15:44:19.221213435Z DEBUG:httpcore.http11:send_request_body.complete
2025-09-24T15:44:19.221241734Z DEBUG:httpcore.http11:receive_response_headers.started request=<Request [b'POST']>
2025-09-24T15:44:19.222983609Z DEBUG:httpcore.http11:receive_response_headers.complete return_value=(b'HTTP/1.1', 200, b'OK', [(b'date', b'Wed, 24 Sep 2025 15:44:19 GMT'), (b'server', b'uvicorn'), (b'content-disposition', b'attachment; filename=speech.mp3'), (b'x-accel-buffering', b'no'), (b'cache-control', b'no-cache'), (b'content-type', b'audio/mpeg'), (b'Transfer-Encoding', b'chunked')])
2025-09-24T15:44:19.223135234Z INFO:httpx:HTTP Request: POST http://kokoro:8880/v1/audio/speech "HTTP/1.1 200 OK"
2025-09-24T15:44:19.223185046Z DEBUG:openai._base_client:HTTP Response: POST http://kokoro:8880/v1/audio/speech "200 OK" Headers({'date': 'Wed, 24 Sep 2025 15:44:19 GMT', 'server': 'uvicorn', 'content-disposition': 'attachment; filename=speech.mp3', 'x-accel-buffering': 'no', 'cache-control': 'no-cache', 'content-type': 'audio/mpeg', 'transfer-encoding': 'chunked'})
2025-09-24T15:44:19.223222908Z DEBUG:openai._base_client:request_id: None
2025-09-24T15:44:19.223282976Z DEBUG:httpcore.http11:receive_response_body.started request=<Request [b'POST']>
2025-09-24T15:44:19.315854968Z DEBUG:httpcore.http11:receive_response_body.complete
2025-09-24T15:44:19.315952104Z DEBUG:httpcore.http11:response_closed.started
2025-09-24T15:44:19.315967689Z DEBUG:httpcore.http11:response_closed.complete
2025-09-24T15:44:19.316087899Z DEBUG:wyoming_openai.handler:Completed synthesis for chunk: From then on, the light never turned off.
2025-09-24T15:44:19.316123931Z DEBUG:wyoming_openai.handler:Failed to parse WAV header: file does not start with RIFF id
2025-09-24T15:44:19.316140883Z DEBUG:wyoming_openai.handler:Could not parse WAV header, using defaults: 24000 Hz
2025-09-24T15:44:19.316848731Z DEBUG:wyoming_openai.handler:Successfully streamed incremental sentence 1, timestamp: 6895.33
2025-09-24T15:44:19.316916232Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:19.316942494Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' the' (length: 4)
2025-09-24T15:44:19.317221059Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'And the'
2025-09-24T15:44:19.345659269Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:19.345683598Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' hallway' (length: 8)
2025-09-24T15:44:19.346040152Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'And the hallway'
2025-09-24T15:44:19.398334583Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:19.398354129Z DEBUG:wyoming_openai.handler:Received synthesis chunk: '?' (length: 1)
2025-09-24T15:44:19.398752127Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'And the hallway?'
2025-09-24T15:44:19.450265664Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:19.450283850Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' It' (length: 3)
2025-09-24T15:44:19.450701728Z INFO:wyoming_openai.handler:Detected 1 ready sentences for immediate synthesis: ['And the hallway?']
2025-09-24T15:44:19.450721340Z DEBUG:wyoming_openai.handler:Starting incremental synthesis for sentence 1: And the hallway?
2025-09-24T15:44:19.451040725Z DEBUG:openai._base_client:Request options: {'method': 'post', 'url': '/audio/speech', 'headers': {'Accept': 'application/octet-stream', 'X-Stainless-Raw-Response': 'stream'}, 'files': None, 'idempotency_key': 'stainless-python-retry-3a78a64c-8358-4bcc-a596-0506cb207821', 'json_data': {'input': 'And the hallway?', 'model': 'kokoro', 'voice': 'af_aoede'}}
2025-09-24T15:44:19.451328667Z DEBUG:openai._base_client:Sending HTTP Request: POST http://kokoro:8880/v1/audio/speech
2025-09-24T15:44:19.451571023Z DEBUG:httpcore.http11:send_request_headers.started request=<Request [b'POST']>
2025-09-24T15:44:19.451807076Z DEBUG:httpcore.http11:send_request_headers.complete
2025-09-24T15:44:19.451829749Z DEBUG:httpcore.http11:send_request_body.started request=<Request [b'POST']>
2025-09-24T15:44:19.451962864Z DEBUG:httpcore.http11:send_request_body.complete
2025-09-24T15:44:19.451986607Z DEBUG:httpcore.http11:receive_response_headers.started request=<Request [b'POST']>
2025-09-24T15:44:19.453881976Z DEBUG:httpcore.http11:receive_response_headers.complete return_value=(b'HTTP/1.1', 200, b'OK', [(b'date', b'Wed, 24 Sep 2025 15:44:19 GMT'), (b'server', b'uvicorn'), (b'content-disposition', b'attachment; filename=speech.mp3'), (b'x-accel-buffering', b'no'), (b'cache-control', b'no-cache'), (b'content-type', b'audio/mpeg'), (b'Transfer-Encoding', b'chunked')])
2025-09-24T15:44:19.454045823Z INFO:httpx:HTTP Request: POST http://kokoro:8880/v1/audio/speech "HTTP/1.1 200 OK"
2025-09-24T15:44:19.454075302Z DEBUG:openai._base_client:HTTP Response: POST http://kokoro:8880/v1/audio/speech "200 OK" Headers({'date': 'Wed, 24 Sep 2025 15:44:19 GMT', 'server': 'uvicorn', 'content-disposition': 'attachment; filename=speech.mp3', 'x-accel-buffering': 'no', 'cache-control': 'no-cache', 'content-type': 'audio/mpeg', 'transfer-encoding': 'chunked'})
2025-09-24T15:44:19.454084879Z DEBUG:openai._base_client:request_id: None
2025-09-24T15:44:19.454153550Z DEBUG:httpcore.http11:receive_response_body.started request=<Request [b'POST']>
2025-09-24T15:44:19.535007977Z DEBUG:httpcore.http11:receive_response_body.complete
2025-09-24T15:44:19.535023106Z DEBUG:httpcore.http11:response_closed.started
2025-09-24T15:44:19.535047727Z DEBUG:httpcore.http11:response_closed.complete
2025-09-24T15:44:19.535127468Z DEBUG:wyoming_openai.handler:Completed synthesis for chunk: And the hallway?
2025-09-24T15:44:19.535143423Z DEBUG:wyoming_openai.handler:Failed to parse WAV header: file does not start with RIFF id
2025-09-24T15:44:19.535154436Z DEBUG:wyoming_openai.handler:Could not parse WAV header, using defaults: 24000 Hz
2025-09-24T15:44:19.535462957Z DEBUG:wyoming_openai.handler:Successfully streamed incremental sentence 1, timestamp: 7312.25
2025-09-24T15:44:19.535497642Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:19.535507754Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' felt' (length: 5)
2025-09-24T15:44:19.535810348Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'It felt'
2025-09-24T15:44:19.569076051Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:19.569090148Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' warmer' (length: 7)
2025-09-24T15:44:19.569363304Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'It felt warmer'
2025-09-24T15:44:19.621482911Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:19.621510552Z DEBUG:wyoming_openai.handler:Received synthesis chunk: '.' (length: 1)
2025-09-24T15:44:19.621828076Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'It felt warmer.'
2025-09-24T15:44:19.675110380Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:19.675131820Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' Even' (length: 5)
2025-09-24T15:44:19.675468494Z INFO:wyoming_openai.handler:Detected 1 ready sentences for immediate synthesis: ['It felt warmer.']
2025-09-24T15:44:19.675481757Z DEBUG:wyoming_openai.handler:Starting incremental synthesis for sentence 1: It felt warmer.
2025-09-24T15:44:19.675739426Z DEBUG:openai._base_client:Request options: {'method': 'post', 'url': '/audio/speech', 'headers': {'Accept': 'application/octet-stream', 'X-Stainless-Raw-Response': 'stream'}, 'files': None, 'idempotency_key': 'stainless-python-retry-3b3786d8-e312-4428-a348-78d2c3f68115', 'json_data': {'input': 'It felt warmer.', 'model': 'kokoro', 'voice': 'af_aoede'}}
2025-09-24T15:44:19.675987091Z DEBUG:openai._base_client:Sending HTTP Request: POST http://kokoro:8880/v1/audio/speech
2025-09-24T15:44:19.676194925Z DEBUG:httpcore.http11:send_request_headers.started request=<Request [b'POST']>
2025-09-24T15:44:19.676350831Z DEBUG:httpcore.http11:send_request_headers.complete
2025-09-24T15:44:19.676357631Z DEBUG:httpcore.http11:send_request_body.started request=<Request [b'POST']>
2025-09-24T15:44:19.676457038Z DEBUG:httpcore.http11:send_request_body.complete
2025-09-24T15:44:19.676463801Z DEBUG:httpcore.http11:receive_response_headers.started request=<Request [b'POST']>
2025-09-24T15:44:19.678137125Z DEBUG:httpcore.http11:receive_response_headers.complete return_value=(b'HTTP/1.1', 200, b'OK', [(b'date', b'Wed, 24 Sep 2025 15:44:19 GMT'), (b'server', b'uvicorn'), (b'content-disposition', b'attachment; filename=speech.mp3'), (b'x-accel-buffering', b'no'), (b'cache-control', b'no-cache'), (b'content-type', b'audio/mpeg'), (b'Transfer-Encoding', b'chunked')])
2025-09-24T15:44:19.678262122Z INFO:httpx:HTTP Request: POST http://kokoro:8880/v1/audio/speech "HTTP/1.1 200 OK"
2025-09-24T15:44:19.678305492Z DEBUG:openai._base_client:HTTP Response: POST http://kokoro:8880/v1/audio/speech "200 OK" Headers({'date': 'Wed, 24 Sep 2025 15:44:19 GMT', 'server': 'uvicorn', 'content-disposition': 'attachment; filename=speech.mp3', 'x-accel-buffering': 'no', 'cache-control': 'no-cache', 'content-type': 'audio/mpeg', 'transfer-encoding': 'chunked'})
2025-09-24T15:44:19.678314254Z DEBUG:openai._base_client:request_id: None
2025-09-24T15:44:19.678400826Z DEBUG:httpcore.http11:receive_response_body.started request=<Request [b'POST']>
2025-09-24T15:44:19.739802549Z DEBUG:httpcore.http11:receive_response_body.complete
2025-09-24T15:44:19.739836441Z DEBUG:httpcore.http11:response_closed.started
2025-09-24T15:44:19.739872847Z DEBUG:httpcore.http11:response_closed.complete
2025-09-24T15:44:19.739939626Z DEBUG:wyoming_openai.handler:Completed synthesis for chunk: It felt warmer.
2025-09-24T15:44:19.739968517Z DEBUG:wyoming_openai.handler:Failed to parse WAV header: file does not start with RIFF id
2025-09-24T15:44:19.739982905Z DEBUG:wyoming_openai.handler:Could not parse WAV header, using defaults: 24000 Hz
2025-09-24T15:44:19.740341039Z DEBUG:wyoming_openai.handler:Successfully streamed incremental sentence 1, timestamp: 7753.17
2025-09-24T15:44:19.741853985Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:19.741872102Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' when' (length: 5)
2025-09-24T15:44:19.742151075Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'Even when'
2025-09-24T15:44:19.794006646Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:19.794030023Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' no' (length: 3)
2025-09-24T15:44:19.794405090Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'Even when no'
2025-09-24T15:44:19.846843035Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:19.846860509Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' one' (length: 4)
2025-09-24T15:44:19.847104797Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'Even when no one'
2025-09-24T15:44:19.898183911Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:19.898209071Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' was' (length: 4)
2025-09-24T15:44:19.898484678Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'Even when no one was'
2025-09-24T15:44:19.951048004Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:19.951065353Z DEBUG:wyoming_openai.handler:Received synthesis chunk: ' home' (length: 5)
2025-09-24T15:44:19.951390753Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'Even when no one was home'
2025-09-24T15:44:20.003986653Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-chunk
2025-09-24T15:44:20.004015437Z DEBUG:wyoming_openai.handler:Received synthesis chunk: '.' (length: 1)
2025-09-24T15:44:20.004472046Z DEBUG:wyoming_openai.handler:No complete sentences ready yet, accumulator has: 'Even when no one was home.'
2025-09-24T15:44:20.057008123Z DEBUG:wyoming_openai.handler:Incoming event type synthesize
2025-09-24T15:44:20.057025539Z DEBUG:wyoming_openai.handler:Handling synthesize request Synthesize(text='In a quiet hallway, a single light flickered once, twice—then stayed on.\n\nIt wasn’t broken. It was watching.\n\nEvery night at 10:47, the shadow of a child would appear on the wall. But no child lived there.\n\nOne evening, the light blinked slowly—once, twice, three times.\n\nAnd the shadow waved back.\n\nFrom then on, the light never turned off. And the hallway? It felt warmer. Even when no one was home.', voice=SynthesizeVoice(name='af_aoede', language=None, speaker=None), context=None)
2025-09-24T15:44:20.057043278Z DEBUG:wyoming_openai.handler:Ignoring standalone synthesize event - streaming synthesis is already active
2025-09-24T15:44:20.057080363Z DEBUG:wyoming_openai.handler:Incoming event type synthesize-stop
2025-09-24T15:44:20.057090850Z INFO:wyoming_openai.handler:Processing final remaining text: 'Even when no one was home.'
2025-09-24T15:44:20.057141133Z DEBUG:wyoming_openai.handler:Starting incremental synthesis for sentence 1: Even when no one was home.
2025-09-24T15:44:20.057440838Z DEBUG:openai._base_client:Request options: {'method': 'post', 'url': '/audio/speech', 'headers': {'Accept': 'application/octet-stream', 'X-Stainless-Raw-Response': 'stream'}, 'files': None, 'idempotency_key': 'stainless-python-retry-1c88b545-75f3-425c-93a1-5ca01c00078d', 'json_data': {'input': 'Even when no one was home.', 'model': 'kokoro', 'voice': 'af_aoede'}}
2025-09-24T15:44:20.057731660Z DEBUG:openai._base_client:Sending HTTP Request: POST http://kokoro:8880/v1/audio/speech
2025-09-24T15:44:20.057972289Z DEBUG:httpcore.http11:send_request_headers.started request=<Request [b'POST']>
2025-09-24T15:44:20.058171974Z DEBUG:httpcore.http11:send_request_headers.complete
2025-09-24T15:44:20.058202644Z DEBUG:httpcore.http11:send_request_body.started request=<Request [b'POST']>
2025-09-24T15:44:20.058315113Z DEBUG:httpcore.http11:send_request_body.complete
2025-09-24T15:44:20.058330869Z DEBUG:httpcore.http11:receive_response_headers.started request=<Request [b'POST']>
2025-09-24T15:44:20.059925192Z DEBUG:httpcore.http11:receive_response_headers.complete return_value=(b'HTTP/1.1', 200, b'OK', [(b'date', b'Wed, 24 Sep 2025 15:44:19 GMT'), (b'server', b'uvicorn'), (b'content-disposition', b'attachment; filename=speech.mp3'), (b'x-accel-buffering', b'no'), (b'cache-control', b'no-cache'), (b'content-type', b'audio/mpeg'), (b'Transfer-Encoding', b'chunked')])
2025-09-24T15:44:20.060026941Z INFO:httpx:HTTP Request: POST http://kokoro:8880/v1/audio/speech "HTTP/1.1 200 OK"
2025-09-24T15:44:20.060070203Z DEBUG:openai._base_client:HTTP Response: POST http://kokoro:8880/v1/audio/speech "200 OK" Headers({'date': 'Wed, 24 Sep 2025 15:44:19 GMT', 'server': 'uvicorn', 'content-disposition': 'attachment; filename=speech.mp3', 'x-accel-buffering': 'no', 'cache-control': 'no-cache', 'content-type': 'audio/mpeg', 'transfer-encoding': 'chunked'})
2025-09-24T15:44:20.060108553Z DEBUG:openai._base_client:request_id: None
2025-09-24T15:44:20.060155364Z DEBUG:httpcore.http11:receive_response_body.started request=<Request [b'POST']>
2025-09-24T15:44:20.134724025Z DEBUG:httpcore.http11:receive_response_body.complete
2025-09-24T15:44:20.134771178Z DEBUG:httpcore.http11:response_closed.started
2025-09-24T15:44:20.134820815Z DEBUG:httpcore.http11:response_closed.complete
2025-09-24T15:44:20.134916595Z DEBUG:wyoming_openai.handler:Completed synthesis for chunk: Even when no one was home.
2025-09-24T15:44:20.134948679Z DEBUG:wyoming_openai.handler:Failed to parse WAV header: file does not start with RIFF id
2025-09-24T15:44:20.134968169Z DEBUG:wyoming_openai.handler:Could not parse WAV header, using defaults: 24000 Hz
2025-09-24T15:44:20.135331494Z DEBUG:wyoming_openai.handler:Successfully streamed incremental sentence 1, timestamp: 8314.08
2025-09-24T15:44:20.135362532Z DEBUG:wyoming_openai.handler:Streaming synthesis completed with text: In a quiet hallway, a single light flickered once, twice—then stayed on.
2025-09-24T15:44:20.135364327Z 
2025-09-24T15:44:20.135365247Z It wasn’t broken. It was w...
2025-09-24T15:44:20.135383003Z DEBUG:wyoming_openai.handler:Outgoing event type audio-stop
2025-09-24T15:44:20.135403719Z DEBUG:wyoming_openai.handler:Outgoing event type synthesize-stopped
2025-09-24T15:44:20.135432828Z INFO:wyoming_openai.handler:Successfully completed incremental streaming synthesis, final timestamp: 8314.08

[K-RAD]

I added the same TTS/STT endpoints to OpenWebui to test out, and there the streaming works as expected

[roryeckel]

I see, looks like a regression in 0.3.7

[roryeckel]

Feel free to try out ghcr.io/roryeckel/wyoming_openai:pr-33. I have some final error handling improvements that I still want to make, so it's a draft for now.