Merge branch 'v1' of github.com:mutablelogic/go-whisper into stream

Author: djthorpe

Makefile

@@ -56,7 +56,7 @@ docker: docker-dep submodule
 		--build-arg OS=${OS} \
 		--build-arg SOURCE=${BUILD_MODULE} \
 		--build-arg VERSION=${VERSION} \
-		-f etc/Dockerfile.${ARCH} .
+		-f etc/Dockerfile.${OS}-${ARCH} .
 
 # Test whisper bindings
 test: generate libwhisper libggml

README.md

@@ -62,7 +62,7 @@ curl -F model=ggml-medium-q5_0 -F file=@samples/jfk.wav localhost:8080/v1/audio/
 To translate a media file into a different language, you can use the following command:
 
 ```bash
-curl -F model=ggml-medium-q5_0 -F file=@samples/ge-podcast.wav -F language=en localhost:8080/v1/audio/translations\?stream=true
+curl -F model=ggml-medium-q5_0 -F file=@samples/de-podcast.wav -F language=en localhost:8080/v1/audio/translations\?stream=true
 ```
 
 There's more information on the API [here](doc/API.md).

cmd/server/flags.go

@@ -20,9 +20,9 @@ func NewFlags(name string, args []string) (*Flags, error) {
 		FlagSet: flag.NewFlagSet(name, flag.ContinueOnError),
 	}
 	flags.endpoint = flags.String("endpoint", "/v1", "HTTP endpoint")
-	flags.listen = flags.String("listen", ":8080", "HTTP Listen address")
+	flags.listen = flags.String("listen", "127.0.0.1:8080", "HTTP Listen address")
 	flags.dir = flags.String("dir", "${WHISPER_DATA}", "Model data directory")
-	flags.debug = flags.Bool("debug", false, "Display debug information")
+	flags.debug = flags.Bool("debug", false, "Output additional debug information")
 
 	// Parse flags and return any error
 	return flags, flags.Parse(args)

cmd/server/main.go

@@ -15,6 +15,7 @@ import (
 	httpserver "github.com/mutablelogic/go-server/pkg/httpserver"
 	whisper "github.com/mutablelogic/go-whisper"
 	api "github.com/mutablelogic/go-whisper/pkg/whisper/api"
+	version "github.com/mutablelogic/go-whisper/pkg/whisper/version"
 )
 
 func main() {
@@ -45,6 +46,16 @@ func main() {
 		os.Exit(-1)
 	}
 
+	// Print version
+	if version.GitSource != "" {
+		log.Println(name, version.GitSource)
+	} else {
+		log.Println(name)
+	}
+	if version.GitTag != "" {
+		log.Println("Version:", version.GitTag)
+	}
+
 	// Create a whisper service
 	log.Println("Storing models at", dir)
 	opts := []whisper.Opt{
@@ -77,7 +88,7 @@ func main() {
 	api.RegisterEndpoints(flags.Endpoint(), mux, whisper)
 
 	// Create a new HTTP server
-	log.Println("List address", flags.Listen())
+	log.Println("Listen address", flags.Listen())
 	server, err := httpserver.Config{
 		Listen: flags.Listen(),
 		Router: mux,

doc/API.md

@@ -144,9 +144,20 @@ event: ok
 
 ### Translation
 
-This is the same as transcription (above) except that the `language` parameter is not optional, and should be the language to translate the audio into.
+This is the same as transcription (above) except that the `language` parameter is always set to 'en', to translate the audio into English.
 
 ```html
 POST /v1/audio/translations
 POST /v1/audio/translations?stream={bool}
 ```
+
+### Diarization
+
+To diarize an Enlgish-language audio file, use the following endpoint:
+
+```html
+POST /v1/audio/diarize
+POST /v1/audio/diarize?stream={bool}
+```
+
+The segments returned include a "speaker_turn" field which indicates that the segment is a new speaker. It requires a separate download of a [diarization model](https://huggingface.co/akashmjn/tinydiarize-whisper.cpp).

etc/Dockerfile.amd64 renamed to etc/Dockerfile.linux-amd64

@@ -8,7 +8,7 @@ require (
 	github.com/djthorpe/go-tablewriter v0.0.8
 	github.com/go-audio/wav v1.1.0
 	github.com/mutablelogic/go-client v1.0.9
-	github.com/mutablelogic/go-media v1.6.10
+	github.com/mutablelogic/go-media v1.6.11
 	github.com/mutablelogic/go-server v1.4.15
 	github.com/stretchr/testify v1.9.0
 )

etc/Dockerfile.arm64 renamed to etc/Dockerfile.linux-arm64

@@ -26,8 +26,8 @@ github.com/mattn/go-runewidth v0.0.16 h1:E5ScNMtiwvlvB5paMFdw9p4kSQzbXFikJ5SQO6T
 github.com/mattn/go-runewidth v0.0.16/go.mod h1:Jdepj2loyihRzMpdS35Xk/zdY8IAYHsh153qUoGf23w=
 github.com/mutablelogic/go-client v1.0.9 h1:Eh4sjQOFDldP/L3IizqkcOD3WigZR+u1VaHTUM4ujYw=
 github.com/mutablelogic/go-client v1.0.9/go.mod h1:VLyB8j8IBJSK/FXvvqhmq93PRWDKkyLu8R7V2Vudb6A=
-github.com/mutablelogic/go-media v1.6.10 h1:LJCNGiAJHFETtATbktTpe38lquUDiLjgeaWXsi1fzI8=
-github.com/mutablelogic/go-media v1.6.10/go.mod h1:HulNT0yyH63a3FRlbuzNDakhOypYrmtFVkHEXZjDgAY=
+github.com/mutablelogic/go-media v1.6.11 h1:czwRvuWIaqDArZrHv0e7nEIjXJkCbnNWkrQzkTOs96w=
+github.com/mutablelogic/go-media v1.6.11/go.mod h1:HulNT0yyH63a3FRlbuzNDakhOypYrmtFVkHEXZjDgAY=
 github.com/mutablelogic/go-server v1.4.15 h1:jOvVdDmVK+PGCMBAk5atKHVonnccwy/b4dWwWFAOTso=
 github.com/mutablelogic/go-server v1.4.15/go.mod h1:9nenPAohKu8bFoRgwHJh+3s8h0kLFjUAb8KZvT1TQNU=
 github.com/pmezard/go-difflib v1.0.0 h1:4DBwDE0NGyQoBHbLQYPwSUPoCMWR5BEzIk/f1lZbAQM=

go.mod

@@ -70,7 +70,7 @@ func RegisterEndpoints(base string, mux *http.ServeMux, whisper *whisper.Whisper
 
 		switch r.Method {
 		case http.MethodPost:
-			TranscribeFile(r.Context(), whisper, w, r, true)
+			TranscribeFile(r.Context(), whisper, w, r, Translate)
 		default:
 			httpresponse.Error(w, http.StatusMethodNotAllowed)
 		}
@@ -84,7 +84,21 @@ func RegisterEndpoints(base string, mux *http.ServeMux, whisper *whisper.Whisper
 
 		switch r.Method {
 		case http.MethodPost:
-			TranscribeFile(r.Context(), whisper, w, r, false)
+			TranscribeFile(r.Context(), whisper, w, r, Transcribe)
+		default:
+			httpresponse.Error(w, http.StatusMethodNotAllowed)
+		}
+	})
+
+	// Diarize: POST /v1/audio/diarize
+	//   Transcribes audio into the input language - language parameter should be set to the source
+	//   language of the audio. Output speaker parts.
+	mux.HandleFunc(joinPath(base, "audio/diarize"), func(w http.ResponseWriter, r *http.Request) {
+		defer r.Body.Close()
+
+		switch r.Method {
+		case http.MethodPost:
+			TranscribeFile(r.Context(), whisper, w, r, Diarize)
 		default:
 			httpresponse.Error(w, http.StatusMethodNotAllowed)
 		}