File size: 4,797 Bytes
95636c5
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
import logging
import os
import uuid
from contextlib import asynccontextmanager
from tempfile import NamedTemporaryFile

import boto3
import torchaudio
from fastapi import BackgroundTasks, Depends, FastAPI, Header, HTTPException
from fastapi.security import APIKeyHeader
from pydantic import BaseModel

from inference import load_models, process_voice_conversion

logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)

# Global variables
models = None
API_KEY = os.getenv("API_KEY")

api_key_header = APIKeyHeader(name="Authorization", auto_error=False)


async def verify_api_key(authorization: str = Header(None)):
    if not authorization:
        logger.warning("No API key provided")
        raise HTTPException(status_code=401, detail="API key is missing")

    if authorization.startswith("Bearer "):
        token = authorization.replace("Bearer ", "")
    else:
        token = authorization

    if token != API_KEY:
        logger.warning("Invalid API key provided")
        raise HTTPException(status_code=401, detail="Invalid API key")

    return token


def get_s3_client():
    client_kwargs = {'region_name': os.getenv("AWS_REGION", "us-east-1")}

    if os.getenv("AWS_ACCESS_KEY_ID") and os.getenv("AWS_SECRET_ACCESS_KEY"):
        client_kwargs.update({
            'aws_access_key_id': os.getenv("AWS_ACCESS_KEY_ID"),
            'aws_secret_access_key': os.getenv("AWS_SECRET_ACCESS_KEY")
        })

    return boto3.client('s3', **client_kwargs)


s3_client = get_s3_client()

S3_PREFIX = os.getenv("S3_PREFIX", "seedvc-outputs")
S3_BUCKET = os.getenv("S3_BUCKET", "elevenlabs-clone")


@asynccontextmanager
async def lifespan(app: FastAPI):
    global models
    logger.info("Loading Seed-VC model...")
    try:
        models = load_models()

        logger.info("Seed-VC model loaded successfully")
    except Exception as e:
        logger.error(f"Failed to load model: {e}")
        raise

    yield

    logger.info("Shutting down Seed-VC API")

app = FastAPI(title="Seed-VC API",
              lifespan=lifespan)

TARGET_VOICES = {
    "andreas": "examples/reference/andreas1.wav",
    "woman": "examples/reference/s1p1.wav",
    "trump": "examples/reference/trump_0.wav",
}


class VoiceConversionRequest(BaseModel):
    source_audio_key: str
    target_voice: str


@app.post("/convert", dependencies=[Depends(verify_api_key)])
async def generate_speech(request: VoiceConversionRequest, background_tasks: BackgroundTasks):
    if not models:
        raise HTTPException(status_code=500, detail="Model not loaded")

    if request.target_voice not in TARGET_VOICES:
        raise HTTPException(
            status_code=400, detail=f"Target voice not supported. Choose from: {', '.join(TARGET_VOICES.keys())}")

    try:
        target_audio_path = TARGET_VOICES[request.target_voice]
        logger.info(
            f"Converting voice: {request.source_audio_key} to {request.target_voice}")

        # Generate a unique filename
        audio_id = str(uuid.uuid4())
        output_filename = f"{audio_id}.wav"
        local_path = f"/tmp/{output_filename}"

        logger.info("Downloading source audio")
        source_temp = NamedTemporaryFile(delete=False, suffix=".wav")
        try:
            s3_client.download_fileobj(
                S3_BUCKET, Key=request.source_audio_key, Fileobj=source_temp)
            source_temp.close()
        except Exception as e:
            os.unlink(source_temp.name)
            raise HTTPException(
                status_code=404, detail="Source audio not found")

        vc_wave, sr = process_voice_conversion(
            models=models, source=source_temp.name, target_name=target_audio_path, output=None)

        os.unlink(source_temp.name)

        torchaudio.save(local_path, vc_wave, sr)

        # Upload to S3
        s3_key = f"{S3_PREFIX}/{output_filename}"
        s3_client.upload_file(local_path, S3_BUCKET, s3_key)

        presigned_url = s3_client.generate_presigned_url(
            'get_object',
            Params={'Bucket': S3_BUCKET, 'Key': s3_key},
            ExpiresIn=3600
        )

        background_tasks.add_task(os.remove, local_path)

        return {
            "audio_url": presigned_url,
            "s3_key": s3_key
        }
    except Exception as e:
        logger.error(f"Error in voice conversion: {e}")
        raise HTTPException(
            status_code=500, detail="Error in voice conversion")


@app.get("/voices", dependencies=[Depends(verify_api_key)])
async def list_voices():
    return {"voices": list(TARGET_VOICES.keys())}


@app.get("/health", dependencies=[Depends(verify_api_key)])
async def health_check():
    if models:
        return {"status": "healthy", "model": "loaded"}
    return {"status": "unhealthy", "model": "not loaded"}