anuj-exe/coqui-test-api

1import os2import time3import shutil4from fastapi import FastAPI, UploadFile, File, Form5from fastapi.responses import FileResponse, JSONResponse6from TTS.api import TTS7 8os.environ["COQUI_TOS_AGREED"] = "1"9 10app = FastAPI()11 12tts = TTS("tts_models/multilingual/multi-dataset/xtts_v2", gpu=False)13 14DEFAULT_SPEAKER = "speakers/speaker_reference.wav"15 16SUPPORTED_LANGUAGES = [17    "en","es","fr","de","it","pt","pl","tr","ru","nl",18    "cs","ar","zh-cn","hu","ko","ja","hi"19]20 21 22@app.post("/generate")23async def generate_audio(24    text: str = Form(...),25    language: str = Form("en"),26    speaker: UploadFile = File(None)27):28    # 🔴 Validate input29    if not text.strip():30        return JSONResponse({"error": "Text is empty"}, status_code=400)31 32    if language not in SUPPORTED_LANGUAGES:33        return JSONResponse({34            "error": f"Unsupported language '{language}'",35            "supported_languages": SUPPORTED_LANGUAGES36        }, status_code=400)37 38    # 🎤 Handle speaker39    speaker_path = None40    speaker_used = "none"41 42    try:43        if speaker:44            speaker_path = f"/tmp/{speaker.filename}"45            with open(speaker_path, "wb") as buffer:46                shutil.copyfileobj(speaker.file, buffer)47            speaker_used = "uploaded"48 49        elif os.path.exists(DEFAULT_SPEAKER):50            speaker_path = DEFAULT_SPEAKER51            speaker_used = "default"52 53        # Unique output54        output_path = f"/tmp/output_{int(time.time())}.wav"55 56        start = time.time()57 58        # 🔥 Critical section (wrap this!)59        tts.tts_to_file(60            text=text,61            speaker_wav=speaker_path,62            language=language,63            file_path=output_path64        )65 66        duration = round(time.time() - start, 2)67 68        return {69            "meta": {70                "time_sec": duration,71                "language": language,72                "speaker": speaker_used73            },74            "audio_url": f"/audio?path={output_path}"75        }76 77    except AssertionError as e:78        return JSONResponse({79            "error": "TTS model assertion error",80            "details": str(e)81        }, status_code=400)82 83    except Exception as e:84        return JSONResponse({85            "error": "Internal server error",86            "details": str(e)87        }, status_code=500)88 89 90@app.get("/audio")91def get_audio(path: str):92    if not os.path.exists(path):93        return JSONResponse({"error": "File not found"}, status_code=404)94 95    return FileResponse(path, media_type="audio/wav", filename="output.wav")