test

Dockerfile

@@ -0,0 +1,16 @@
+# Read the doc: https://huggingface.co/docs/hub/spaces-sdks-docker
+# you will also find guides on how best to write your Dockerfile
+
+FROM python:3.9
+
+RUN useradd -m -u 1000 user
+USER user
+ENV PATH="/home/user/.local/bin:$PATH"
+
+WORKDIR /app
+
+COPY --chown=user ./requirements.txt requirements.txt
+RUN pip install --no-cache-dir --upgrade -r requirements.txt
+
+COPY --chown=user . /app
+CMD ["uvicorn", "app:app", "--host", "0.0.0.0", "--port", "7860"]

app.py

@@ -1,56 +1,52 @@
-from flask import Flask, request, jsonify
+from fastapi import FastAPI, File, UploadFile, HTTPException
 from transformers import AutoProcessor, SeamlessM4Tv2Model
 import numpy as np
 import wave
 import os
-from huggingface_hub import InferenceClient, login
-from dotenv import load_dotenv
-app = Flask(__name__)
+from starlette.responses import JSONResponse
 
- 
+app = FastAPI()
 
-processor = AutoProcessor.from_pretrained("facebook/seamless-m4t-v2-large" )
+processor = AutoProcessor.from_pretrained("facebook/seamless-m4t-v2-large")
 model = SeamlessM4Tv2Model.from_pretrained("facebook/seamless-m4t-v2-large")
 
 UPLOAD_FOLDER = "audio_files"
 os.makedirs(UPLOAD_FOLDER, exist_ok=True)
 
-@app.route("/", methods=["GET"])
+@app.get("/")
 def return_text():
-    return jsonify({"text": "Hello, world!"})
+    return {"text": "Hello, world!"}
 
-@app.route("/record", methods=["POST"])
-def record_audio():
-    file = request.files['audio']
-    filename = os.path.join(UPLOAD_FOLDER, file.filename)
-    file.save(filename)
+@app.post("/record")
+async def record_audio(audio: UploadFile = File(...)):
+    filename = os.path.join(UPLOAD_FOLDER, audio.filename)
+    with open(filename, "wb") as buffer:
+        buffer.write(await audio.read())
     
-    # Charger et traiter l'audio
-    audio_data, orig_freq = torchaudio.load(filename)
-    audio_inputs = processor(audios=audio_data, return_tensors="pt")
-    output_tokens = model.generate(**audio_inputs, tgt_lang="fra", generate_speech=False)
-    translated_text = processor.decode(output_tokens[0].tolist()[0], skip_special_tokens=True)
+    try:
+        # Charger et traiter l'audio
+        audio_data, orig_freq = torchaudio.load(filename)
+        audio_inputs = processor(audios=audio_data, return_tensors="pt")
+        output_tokens = model.generate(**audio_inputs, tgt_lang="fra", generate_speech=False)
+        translated_text = processor.decode(output_tokens[0].tolist()[0], skip_special_tokens=True)
+    except Exception as e:
+        raise HTTPException(status_code=500, detail=f"Erreur de transcription: {str(e)}")
     
-    return jsonify({"translated_text": translated_text})
+    return JSONResponse(content={"translated_text": translated_text})
 
-@app.route("/text_to_speech", methods=["POST"])
-def text_to_speech():
-    data = request.get_json()
-    text = data.get("text")
-    src_lang = data.get("src_lang")
-    tgt_lang = data.get("tgt_lang")
+@app.post("/text_to_speech")
+async def text_to_speech(text: str, src_lang: str, tgt_lang: str):
+    try:
+        text_inputs = processor(text=text, src_lang=src_lang, return_tensors="pt")
+        audio_array = model.generate(**text_inputs, tgt_lang=tgt_lang)[0].cpu().numpy().squeeze()
     
-    text_inputs = processor(text=text, src_lang=src_lang, return_tensors="pt")
-    audio_array = model.generate(**text_inputs, tgt_lang=tgt_lang)[0].cpu().numpy().squeeze()
+        output_filename = os.path.join(UPLOAD_FOLDER, "output.wav")
+        with wave.open(output_filename, "wb") as wf:
+            wf.setnchannels(1)
+            wf.setsampwidth(2)
+            wf.setframerate(16000)
+            wf.writeframes((audio_array * 32767).astype(np.int16).tobytes())
+    except Exception as e:
+        raise HTTPException(status_code=500, detail=f"Erreur de synthèse vocale: {str(e)}")
     
-    output_filename = os.path.join(UPLOAD_FOLDER, "output.wav")
-    with wave.open(output_filename, "wb") as wf:
-        wf.setnchannels(1)
-        wf.setsampwidth(2)
-        wf.setframerate(16000)
-        wf.writeframes((audio_array * 32767).astype(np.int16).tobytes())
-    
-    return jsonify({"audio_url": output_filename})
-
-if __name__ == "__main__":
-    app.run(debug=True)
+    return JSONResponse(content={"audio_url": output_filename})

requirements.txt

@@ -1,4 +1,5 @@
-flask
+fastapi
+uvicorn[standard]
 torch
 numpy
 transformers