这有帮助吗？
# 通过 SSH 或 Jupyter 终端连接
pip install torch torchaudio transformers accelerate
# 安装 Voxtral TTS 包
pip install voxtral-tts
# 或直接使用 HuggingFace
pip install huggingface_hub
huggingface-cli download mistralai/Voxtral-TTS --local-dir ./voxtral-ttsfrom voxtral import VoxtralTTS
# 初始化模型（自动下载约 6 GB 权重）
model = VoxtralTTS.from_pretrained("mistralai/Voxtral-TTS")
model.to("cuda")
# 基础合成
audio = model.synthesize(
text="欢迎来到 Clore.ai——去中心化 GPU 市场。",
language="en"
)
audio.save("output.wav")
print(f"已生成 {audio.duration:.1f} 秒音频")# 从 3 秒参考音频克隆声音
audio = model.synthesize(
text="这是我克隆的声音，在谈论 GPU 计算。",
reference_audio="reference_speaker.wav", # 3 秒以上
language="en"
)
audio.save("cloned_output.wav")# 使用 9 种受支持语言进行合成
languages = {
"en": "Hello, this is Voxtral speaking in English.",
"fr": "Bonjour, c'est Voxtral qui parle en français.",
"de": "Hallo, hier spricht Voxtral auf Deutsch.",
"es": "Hola, Voxtral hablando en español.",
"hi": "नमस्ते, यह Voxtral हिंदी में बोल रहा है।",
"ar": "مرحبا، هذا Voxtral يتحدث بالعربية.",
"pt": "Olá, aqui é o Voxtral falando em português.",
"it": "Ciao, qui parla Voxtral in italiano.",
"ja": "こんにちは、Voxtralが日本語で話しています。",
}
for lang, text in languages.items():
audio = model.synthesize(text=text, language=lang)
audio.save(f"voxtral_{lang}.wav")
print(f"[{lang}] 已生成 {audio.duration:.1f} 秒")# server.py — Voxtral TTS 的 FastAPI 封装
from fastapi import FastAPI, UploadFile, File
from fastapi.responses import StreamingResponse
from voxtral import VoxtralTTS
import io
import soundfile as sf
app = FastAPI(title="Voxtral TTS API")
model = VoxtralTTS.from_pretrained("mistralai/Voxtral-TTS").to("cuda")
@app.post("/synthesize")
async def synthesize(
text: str,
language: str = "en",
reference: UploadFile = File(None)
):
kwargs = {"text": text, "language": language}
if reference:
ref_bytes = await reference.read()
kwargs["reference_audio"] = ref_bytes
audio = model.synthesize(**kwargs)
# 以 WAV 流形式返回
buffer = io.BytesIO()
sf.write(buffer, audio.numpy(), samplerate=24000, format="WAV")
buffer.seek(0)
return StreamingResponse(buffer, media_type="audio/wav")
@app.get("/health")
async def health():
return {"status": "ok", "model": "voxtral-tts", "languages": 9}# 运行 API 服务器
pip install fastapi uvicorn python-multipart soundfile
uvicorn server:app --host 0.0.0.0 --port 8000
# 测试它
curl -X POST "http://localhost:8000/synthesize?text=Hello%20world&language=en" \
--output hello.wavFROM pytorch/pytorch:2.4.0-cuda12.4-cudnn9-devel
WORKDIR /app
RUN pip install voxtral-tts fastapi uvicorn python-multipart soundfile
# 预先下载模型权重
RUN python -c "from voxtral import VoxtralTTS; VoxtralTTS.from_pretrained('mistralai/Voxtral-TTS')"
COPY server.py .
EXPOSE 8000
CMD ["uvicorn", "server:app", "--host", "0.0.0.0", "--port", "8000"]# 构建并运行
docker build -t voxtral-tts-api .
docker run --gpus all -p 8000:8000 voxtral-tts-apiimport concurrent.futures
from voxtral import VoxtralTTS
model = VoxtralTTS.from_pretrained("mistralai/Voxtral-TTS").to("cuda")
# 处理整章有声书
paragraphs = [
"第 1 章：开始……",
"那是一个黑暗而暴风雨肆虐的夜晚……",
"主角向前迈出一步……",
# ... 数百个段落
]
def process_paragraph(idx_text):
idx, text = idx_text
audio = model.synthesize(text=text, language="en")
audio.save(f"chapter1_part{idx:04d}.wav")
return idx
# 顺序处理（受 GPU 限制）
for i, text in enumerate(paragraphs):
process_paragraph((i, text))
print(f"已处理 {len(paragraphs)} 个段落")# 面向直播/实时应用的流式合成
async def stream_synthesis(text: str, language: str = "en"):
"""以流式分块生成音频，以实现低延迟播放。"""
model = VoxtralTTS.from_pretrained("mistralai/Voxtral-TTS").to("cuda")
async for chunk in model.synthesize_stream(
text=text,
language=language,
chunk_size=4096 # 在 24kHz 下每块约 170ms
):
yield chunk.numpy().tobytes()