Update app.py

app.py

@@ -90,7 +90,8 @@ def detect_onnx_models(path):
     else:
         return None
 
-@app.post("/synthesize")
+#@app.post("/synthesize")
+@app.post("/synthesize", response_class=FileResponse)
 async def main(
     request: Request,
     text_input: str = Form(...),
@@ -128,22 +129,9 @@ async def main(
             speaker_selection.options = config["speaker_id_map"].values()
             speaker_selection.layout.visibility = 'visible'
             preview_sid = 0
-            if enhanced_accessibility:
-                playaudio("multispeaker")
         else:
             speaker_selection.layout.visibility = 'hidden'
             preview_sid = None
-
-        if enhanced_accessibility:
-            inferencing(
-                model,
-                config,
-                preview_sid,
-                lan.translate(
-                    config["espeak"]["voice"][:2],
-                    "Interface openned. Write your texts, configure the different synthesis options or download all the voices you want. Enjoy!"
-                )
-            )
     else:
         voice_model_names = []
         for current in onnx_models:
@@ -199,8 +187,11 @@ async def main(
         auto_play = play.value
         inferencing(model, config, sid, text, rate, noise_scale, noise_scale_w, auto_play)
 
-    
-    return {"message": f"Text to synthesize: {text_input}, Speed: {speed_slider}, Play: {play}"}
+        # Save the audio as a temporary WAV file
+    temp_audio_path = os.path.join(tempfile.gettempdir(), "generated_audio.wav")
+    sf.write(temp_audio_path, merged_audio, config["audio"]["sample_rate"])
+    return FileResponse(temp_audio_path)
+#    return {"message": f"Text to synthesize: {text_input}, Speed: {speed_slider}, Play: {play}"}
 
 def load_onnx(model, sess_options, providers = ["CPUExecutionProvider"]):
     _LOGGER.debug("Loading model from %s", model)
@@ -249,8 +240,8 @@ def phonemes_to_ids(config, phonemes: List[str]) -> List[int]:
         ids.extend(id_map[PAD])
     ids.extend(id_map[EOS])
     return ids
-@app.get("/generate_audio")
-async def inferencing(model, config, sid, line, length_scale = 1, noise_scale = 0.667, noise_scale_w = 0.8, auto_play=True):
+
+def inferencing(model, config, sid, line, length_scale = 1, noise_scale = 0.667, noise_scale_w = 0.8, auto_play=True):
     audios = []
     if config["phoneme_type"] == "PhonemeType.ESPEAK":
         config["phoneme_type"] = "espeak"
@@ -284,12 +275,8 @@ async def inferencing(model, config, sid, line, length_scale = 1, noise_scale =
         audios.append(audio)
     merged_audio = np.concatenate(audios)
     sample_rate = config["audio"]["sample_rate"]
-    # Save the audio as a temporary WAV file
-    temp_audio_path = os.path.join(tempfile.gettempdir(), "generated_audio.wav")
-    sf.write(temp_audio_path, merged_audio, config["audio"]["sample_rate"])
     
     # Return the audio file as a FastAPI response
-    return FileResponse(temp_audio_path)
   #  display(Markdown(f"{line}"))
    # display(Audio(merged_audio, rate=sample_rate, autoplay=auto_play))
 
@@ -395,7 +382,7 @@ async def read_root(request: Request):
 
 
 if __name__ == "__main__":
-    main()
+ #   main()
     import uvicorn
     uvicorn.run(app, host="0.0.0.0", port=7860)
 #    main()