Initial FastAPI deployment for MMS speech-to-text on HF Spaces

.gitignore

@@ -0,0 +1,16 @@
+__pycache__/
+*.py[cod]
+*.pyo
+.env
+.venv/
+venv/
+*.egg-info/
+dist/
+build/
+.DS_Store
+*.ipynb_checkpoints/
+.ipynb_checkpoints/
+*.mp3
+*.wav
+*.flac
+*.ogg

Dockerfile

@@ -0,0 +1,18 @@
+FROM python:3.10-slim
+
+RUN apt-get update && apt-get install -y \
+    ffmpeg \
+    libsndfile1 \
+    git \
+    && rm -rf /var/lib/apt/lists/*
+
+WORKDIR /app
+
+COPY requirements.txt .
+RUN pip install --no-cache-dir -r requirements.txt
+
+COPY app.py .
+
+EXPOSE 7860
+
+CMD ["uvicorn", "app:app", "--host", "0.0.0.0", "--port", "7860"]

README.md

@@ -0,0 +1,42 @@
+---
+title: MMS ProxyAPI
+emoji: 🎤
+colorFrom: blue
+colorTo: purple
+sdk: docker
+pinned: false
+---
+
+# MMS Speech-to-Text API
+
+REST API built with FastAPI using the `facebook/mms-1b-all` model — supports 1000+ languages including Twi.
+
+## Endpoints
+
+| Method | Path | Description |
+|--------|------|-------------|
+| GET | `/` | API info |
+| GET | `/health` | Health check |
+| POST | `/transcribe` | Transcribe audio file |
+
+## Usage
+
+```bash
+curl -X POST "https://fredyhoundayi-mms-proxyapi.hf.space/transcribe" \
+  -F "file=@audio.mp3" \
+  -F "language=twi"
+```
+
+### Response
+
+```json
+{
+  "transcription": "awaci ciri duya bi hon",
+  "language": "twi"
+}
+```
+
+## Supported languages
+
+Pass any MMS language code via the `language` field (default: `twi`).
+Full list: https://dl.fbaipublicfiles.com/mms/misc/language_coverage_mms.html

app.py

@@ -0,0 +1,56 @@
+import io
+
+import torch
+import librosa
+from fastapi import FastAPI, File, UploadFile, HTTPException
+from fastapi.responses import JSONResponse
+from transformers import Wav2Vec2ForCTC, AutoProcessor
+
+app = FastAPI(title="MMS Speech-to-Text API", version="1.0.0")
+
+MODEL_ID = "facebook/mms-1b-all"
+processor = None
+model = None
+
+
+@app.on_event("startup")
+async def load_model():
+    global processor, model
+    print("Loading MMS model...")
+    processor = AutoProcessor.from_pretrained(MODEL_ID)
+    model = Wav2Vec2ForCTC.from_pretrained(MODEL_ID)
+    model.eval()
+    print("Model loaded.")
+
+
+@app.get("/")
+def root():
+    return {"message": "MMS Speech-to-Text API", "model": MODEL_ID}
+
+
+@app.get("/health")
+def health():
+    return {"status": "ok", "model_loaded": model is not None}
+
+
+@app.post("/transcribe")
+async def transcribe(file: UploadFile = File(...)):
+    if model is None or processor is None:
+        raise HTTPException(status_code=503, detail="Model not loaded yet")
+
+    audio_bytes = await file.read()
+
+    try:
+        audio, sampling_rate = librosa.load(io.BytesIO(audio_bytes), sr=16000, mono=True)
+    except Exception as e:
+        raise HTTPException(status_code=400, detail=f"Failed to load audio: {e}")
+
+    inputs = processor(audio, sampling_rate=sampling_rate, return_tensors="pt")
+
+    with torch.no_grad():
+        logits = model(**inputs).logits
+
+    predicted_ids = torch.argmax(logits, dim=-1)[0]
+    transcription = processor.decode(predicted_ids)
+
+    return JSONResponse({"transcription": transcription})

requirements.txt

@@ -0,0 +1,10 @@
+fastapi==0.115.0
+uvicorn[standard]==0.30.6
+python-multipart==0.0.9
+torch==2.3.1
+torchaudio==2.3.1
+transformers==4.44.2
+accelerate==0.33.0
+librosa==0.10.2
+numpy==1.26.4
+soundfile==0.12.1