可以api调用吗？

[Nicolas-XF]

可以api调用吗？

[exp555]

import os
import torch
import soundfile as sf
import logging
import argparse
from fastapi import FastAPI, HTTPException
from fastapi.responses import FileResponse
from pydantic import BaseModel
import uvicorn

from datetime import datetime
from cli.SparkTTS import SparkTTS # 确保从正确路径导入
from sparktts.utils.token_parser import LEVELS_MAP_UI

初始化FastAPI应用

app = FastAPI(title="SparkTTS API")

class TTSRequest(BaseModel):
text: str
gender: str = "male"
pitch: int = 3
speed: int = 3

def initialize_model(model_dir="pretrained_models/Spark-TTS-0.5B", device=0):
"""模型初始化（与原代码保持一致）"""
logging.info(f"Loading model from: {model_dir}")

if torch.cuda.is_available():
    device = torch.device(f"cuda:{device}")
elif platform.system() == "Darwin":
    device = torch.device(f"mps:{device}")
else:
    device = torch.device("cpu")

model = SparkTTS(model_dir, device)
return model

@app.post("/generate_voice/")
async def generate_voice(request: TTSRequest):
"""API端点处理语音生成请求"""
try:
# 参数验证
if request.gender.lower() not in ["male", "female"]:
raise ValueError("Invalid gender value")
if not (1 <= request.pitch <= 5) or not (1 <= request.speed <= 5):
raise ValueError("Pitch/Speed must be between 1-5")

    # 转换参数
    pitch_val = LEVELS_MAP_UI[request.pitch]
    speed_val = LEVELS_MAP_UI[request.speed]

    # 生成音频
    save_dir = "api_results"
    os.makedirs(save_dir, exist_ok=True)
    timestamp = datetime.now().strftime("%Y%m%d%H%M%S")
    save_path = os.path.join(save_dir, f"{timestamp}.wav")

    with torch.no_grad():
        wav = model.inference(
            request.text,
            None,  # prompt_speech
            None,  # prompt_text
            request.gender,
            pitch_val,
            speed_val,
        )
        sf.write(save_path, wav, samplerate=16000)

    # 返回音频文件（自动清理需要额外实现）
    return FileResponse(
        save_path,
        media_type="audio/wav",
        headers={"Content-Disposition": f"attachment; filename={timestamp}.wav"}
    )

except Exception as e:
    logging.error(f"Generation failed: {str(e)}")
    raise HTTPException(status_code=400, detail=str(e))

if name == "main":
# 解析命令行参数
parser = argparse.ArgumentParser()
parser.add_argument("--model_dir", type=str, default="pretrained_models/Spark-TTS-0.5B")
parser.add_argument("--device", type=int, default=0)
parser.add_argument("--host", type=str, default="0.0.0.0")
parser.add_argument("--port", type=int, default=7861)
args = parser.parse_args()

# 全局初始化模型
global model
model = initialize_model(args.model_dir, args.device)

# 启动API服务
uvicorn.run(
    app,
    host=args.host,
    port=args.port,
    log_level="info"
)

[breakstring]

你也可以拉我的那个 PR 里的代码（#40 ）试试看。。。这个 Repo 估计就是为了发发论文，感觉根本没人维护。。。也没看到啥人回答问题，PR 也没人评审😮‍💨

[jiakechong1991]

你也可以拉我的那个 PR 里的代码（#40 ）试试看。。。这个 Repo 估计就是为了发发论文，感觉根本没人维护。。。也没看到啥人回答问题，PR 也没人评审😮‍💨

大佬，你要是能提供一个流式接口就好了