ExternalVendorCode/LocalAI
1
0
Fork 0
mirror of https://github.com/mudler/LocalAI.git synced 2025-06-17 22:48:08 +00:00
Code Issues Actions 23 Packages Projects Releases Wiki Activity

feat(tts): add Elevenlabs and OpenAI TTS compatibility layer (#1834)

Browse Source 
* feat(elevenlabs): map elevenlabs API support to TTS

This allows elevenlabs Clients to work automatically with LocalAI by
supporting the elevenlabs API.

The elevenlabs server endpoint is implemented such as it is wired to the
TTS endpoints.

Fixes: https://github.com/mudler/LocalAI/issues/1809

* feat(openai/tts): compat layer with openai tts

Fixes: #1276

* fix: adapt tts CLI
This commit is contained in:
Ettore Di Giacinto
2024-03-14 23:08:34 +01:00
committed by GitHub
parent 45d520f913
commit 20136ca8b7
24 changed files with 454 additions and 338 deletions
Download Patch File Download Diff File Expand all files Collapse all files

22
core/backend/tts.go
View File

@ -29,7 +29,7 @@ func generateUniqueFileName(dir, baseName, ext string) string {
}
}
func ModelTTS(backend, text, modelFile string, loader *model.ModelLoader, appConfig *config.ApplicationConfig, backendConfig config.BackendConfig) (string, *proto.Result, error) {
func ModelTTS(backend, text, modelFile, voice string, loader *model.ModelLoader, appConfig *config.ApplicationConfig, backendConfig config.BackendConfig) (string, *proto.Result, error) {
bb := backend
if bb == "" {
bb = model.PiperBackend
@ -44,12 +44,12 @@ func ModelTTS(backend, text, modelFile string, loader *model.ModelLoader, appCon
model.WithAssetDir(appConfig.AssetsDestination),
model.WithLoadGRPCLoadModelOpts(grpcOpts),
})
piperModel, err := loader.BackendLoader(opts...)
ttsModel, err := loader.BackendLoader(opts...)
if err != nil {
return "", nil, err
}
if piperModel == nil {
if ttsModel == nil {
return "", nil, fmt.Errorf("could not load piper model")
}
@ -57,25 +57,31 @@ func ModelTTS(backend, text, modelFile string, loader *model.ModelLoader, appCon
return "", nil, fmt.Errorf("failed creating audio directory: %s", err)
}
fileName := generateUniqueFileName(appConfig.AudioDir, "piper", ".wav")
fileName := generateUniqueFileName(appConfig.AudioDir, "tts", ".wav")
filePath := filepath.Join(appConfig.AudioDir, fileName)
// If the model file is not empty, we pass it joined with the model path
modelPath := ""
if modelFile != "" {
if bb != model.TransformersMusicGen {
modelPath = filepath.Join(loader.ModelPath, modelFile)
if err := utils.VerifyPath(modelPath, appConfig.ModelPath); err != nil {
// If the model file is not empty, we pass it joined with the model path
// Checking first that it exists and is not outside ModelPath
// TODO: we should actually first check if the modelFile is looking like
// a FS path
mp := filepath.Join(loader.ModelPath, modelFile)
if _, err := os.Stat(mp); err == nil {
if err := utils.VerifyPath(mp, appConfig.ModelPath); err != nil {
return "", nil, err
}
modelPath = mp
} else {
modelPath = modelFile
}
}
res, err := piperModel.TTS(context.Background(), &proto.TTSRequest{
res, err := ttsModel.TTS(context.Background(), &proto.TTSRequest{
Text: text,
Model: modelPath,
Voice: voice,
Dst: filePath,
})