We present LFM2-Audio-1.5B, Liquid AI's first end-to-end audio foundation model. Built with low-latency in mind, the lightweight LFM2 backbone enables real time speech-to-speech conversations without sacrificing quality.
LFM2-Audio supports two generation modes, interleaved and sequential, to maximize performance and quality across different tasks. Interleaved generation outputs text and audio tokens in a fixed interleaved pattern. This approach minimizes time to first audio output and number of tokens generated, making it ideal for naturally flowing real-time speech-to-speech interactions on resource constrained devices. Sequential generation mode, where the model decides when to switch modalities via special tokens, is suitable for non-conversational tasks, such as speech-to-text (ASR) or text-to-speech (TTS).
The package can be installed via pip
pip install liquid-audio
pip install "liquid-audio [demo]" # optional, to install demo dependencies
pip install flash-attn --no-build-isolation # optional, to use flash attention 2. Will fallback to torch SDPA if not installedGeneration is handled by two generation modes, interleaved and sequential, accessible from the methods LFM2AudioModel.generate_interleaved and LFM2AudioModel.generate_sequential respectively. Both are generators that yield torch.Tensors. Text tokens are represented by tensors with 1 entry, and audio tokens are tensors with 8 entries, corresponding to 8 Mimi codebooks.
The LFM2AudioModel class operates on tokens only. The LFM2AudioProcessor class is used convert between tokens and data. For text, this means the conversion from string to tokens and back. For audio inputs, this handles the conversion of waveforms to log-mel features, and for audio outputs, this handles the detokenization of audio tokens to waveform.
To facilitate the creation of inputs for the generation methods and to apply the correct chat templates, use the ChatState helper class. See examples below for usage instructions.
To use the demo interface, make sure to install the extra dependencies in the [demo] group, e.g.
pip install "liquid-audio [demo]"To launch the demo, use the command liquid-audio-demo on the terminal. The demo interface will be available via the url http://localhost:7860.
For multi-turn chat with text and audio output, we use interleaved generation. The system prompt should be set to Respond with interleaved text and audio.. Here we use audio as the first user turn, and text as the second one
Conversation transcript
User
question.mp4
Assistant
Sure! How about "Handcrafted Woodworking, Precision Made for You"? Another option could be "Quality Woodworking, Quality Results." If you want something more personal, you might try "Your Woodworking Needs, Our Expertise."
answer1.mp4
User
My business specialized in chairs, can you give me something related to that?
Assistant
Sure thing! How about “Comfortable Chairs, Crafted with Care” or “Elegant Seats, Handcrafted for You”? Let me know if you’d like a few more options.
answer2.mp4
import torch
import torchaudio
from liquid_audio import LFM2AudioModel, LFM2AudioProcessor, ChatState, LFMModality
# Load models
HF_REPO = "LiquidAI/LFM2-Audio-1.5B"
processor = LFM2AudioProcessor.from_pretrained(HF_REPO).eval()
model = LFM2AudioModel.from_pretrained(HF_REPO).eval()
# Set up inputs for the model
chat = ChatState(processor)
chat.new_turn("system")
chat.add_text("Respond with interleaved text and audio.")
chat.end_turn()
chat.new_turn("user")
wav, sampling_rate = torchaudio.load("assets/question.wav")
chat.add_audio(wav, sampling_rate)
chat.end_turn()
chat.new_turn("assistant")
# Generate text and audio tokens.
text_out: list[torch.Tensor] = []
audio_out: list[torch.Tensor] = []
modality_out: list[LFMModality] = []
for t in model.generate_interleaved(**chat, max_new_tokens=512, audio_temperature=1.0, audio_top_k=4):
if t.numel() == 1:
print(processor.text.decode(t), end="", flush=True)
text_out.append(t)
modality_out.append(LFMModality.TEXT)
else:
audio_out.append(t)
modality_out.append(LFMModality.AUDIO_OUT)
# output: Sure! How about "Handcrafted Woodworking, Precision Made for You"? Another option could be "Quality Woodworking, Quality Results." If you want something more personal, you might try "Your Woodworking Needs, Our Expertise."
# Detokenize audio, removing the last "end-of-audio" codes
# Mimi returns audio at 24kHz
mimi_codes = torch.stack(audio_out[:-1], 1).unsqueeze(0)
with torch.no_grad():
waveform = processor.mimi.decode(mimi_codes)[0]
torchaudio.save("answer1.wav", waveform.cpu(), 24_000)
# Append newly generated tokens to chat history
chat.append(
text = torch.stack(text_out, 1),
audio_out = torch.stack(audio_out, 1),
modality_flag = torch.tensor(modality_out),
)
chat.end_turn()
# Start new turn
chat.new_turn("user")
chat.add_text("My business specialized in chairs, can you give me something related to that?")
chat.end_turn()
chat.new_turn("assistant")
# Generate second turn text and audio tokens.
audio_out: list[torch.Tensor] = []
for t in model.generate_interleaved(**chat, max_new_tokens=512, audio_temperature=1.0, audio_top_k=4):
if t.numel() == 1:
print(processor.text.decode(t), end="", flush=True)
else:
audio_out.append(t)
# output: Sure thing! How about “Comfortable Chairs, Crafted with Care” or “Elegant Seats, Handcrafted for You”? Let me know if you’d like a few more options.
# Detokenize second turn audio, removing the last "end-of-audio" codes
mimi_codes = torch.stack(audio_out[:-1], 1).unsqueeze(0)
with torch.no_grad():
waveform = processor.mimi.decode(mimi_codes)[0]
torchaudio.save("answer2.wav", waveform.cpu(), 24_000)For ASR, we use sequential generation, with the fixed system prompt Perform ASR.. The output is capitalized and punctuated.
Input audio snippet
asr.mp4
Model output: The stale smell of old beer lingers. It takes heat to bring out the odor. A cold dip restores health and zest. A salt pickle tastes fine with ham. Tacos al pastor are my favorite. A zestful food is the hot cross bun.
import torch
import torchaudio
from liquid_audio import LFM2AudioModel, LFM2AudioProcessor, ChatState, LFMModality
# Load models
HF_REPO = "LiquidAI/LFM2-Audio-1.5B"
processor = LFM2AudioProcessor.from_pretrained(HF_REPO).eval()
model = LFM2AudioModel.from_pretrained(HF_REPO).eval()
# Set up inputs for the model
chat = ChatState(processor)
chat.new_turn("system")
chat.add_text("Perform ASR.")
chat.end_turn()
chat.new_turn("user")
wav, sampling_rate = torchaudio.load("assets/asr.wav")
chat.add_audio(wav, sampling_rate)
chat.end_turn()
chat.new_turn("assistant")
# Generate text
for t in model.generate_sequential(**chat, max_new_tokens=512):
if t.numel() == 1:
print(processor.text.decode(t), end="", flush=True)
# Output: The stale smell of old beer lingers. It takes heat to bring out the odor. A cold dip restores health and zest. A salt pickle tastes fine with ham. Tacos al pastor are my favorite. A zestful food is the hot cross bun.For TTS, we also use sequential generation, with the fixed system prompt Perform TTS.. In addition, we can prompt the voice and a style using a natural language description.
TTS Sample
Voice description: A male speaker delivers his lines with a low-pitched voice and an animated tone. The recording is of excellent quality with almost no noise and a very close-sounding atmosphere.
Input sentence: What is this obsession people have with books? They put them in their houses—like they're trophies. What do you need it for after you read it?
Output audio
tts.mp4
import torch
import torchaudio
from liquid_audio import LFM2AudioModel, LFM2AudioProcessor, ChatState, LFMModality
# Load models
HF_REPO = "LiquidAI/LFM2-Audio-1.5B"
processor = LFM2AudioProcessor.from_pretrained(HF_REPO).eval()
model = LFM2AudioModel.from_pretrained(HF_REPO).eval()
# Set up inputs for the model
chat = ChatState(processor)
chat.new_turn("system")
chat.add_text("Perform TTS.\nUse the following voice: A male speaker delivers his lines with a low-pitched voice and an animated tone. The recording is of excellent quality with almost no noise and a very close-sounding atmosphere.")
chat.end_turn()
chat.new_turn("user")
chat.add_text("What is this obsession people have with books? They put them in their houses—like they're trophies. What do you need it for after you read it?")
chat.end_turn()
chat.new_turn("assistant")
# Generate text
audio_out: list[torch.Tensor] = []
for t in model.generate_sequential(**chat, max_new_tokens=512, audio_temperature = 0.8, audio_top_k=64):
if t.numel() > 1:
audio_out.append(t)
# Detokenize audio
mimi_codes = torch.stack(audio_out[:-1], 1).unsqueeze(0)
with torch.no_grad():
waveform = processor.mimi.decode(mimi_codes)[0]
torchaudio.save("tts.wav", waveform.cpu(), 24_000)The code in this repository and associated weights are licensed under the LFM Open License v1.0.
The code for the audio encoder is based on Nvidia NeMo, licensed under Apache 2.0, and the canary-180m-flash checkpoint, licensed under CC-BY 4.0. To simplify dependency resolution, we also ship the Python code of Kyutai Mimi, licensed under the MIT License.