2.1 KiB
2.1 KiB
TTS小模型
我们创建一个webserver, 参考这里提供的安装和代码
import soundfile as sf
from dia.model import Dia
model = Dia.from_pretrained("nari-labs/Dia-1.6B")
text = "[S1] Dia is an open weights text to dialogue model. [S2] You get full control over scripts and voices. [S1] Wow. Amazing. (laughs) [S2] Try it now on Git hub or Hugging Face."
output = model.generate(text)
sf.write("simple.mp3", output, 44100)
上述代码有三个问题
-
output的类型是np.int64,不被audio格式支持
-
sf.write不支持mp3格式,需要换成wav
-
生成的wav文件没有语音
目前1,2,问题解决,第三个问题依然存在
依赖
模型
-
nari-labs/dia-1.6b 下载地址:https://huggingface.com/nari-labs/dia-1.6b
-
dac/weights.pth 下载地址:i
- https://github.com/descriptinc/descript-audio-codec/releases/download/0.0.1/weights.pth
- https://github.com/descriptinc/descript-audio-codec/releases/download/0.0.4/weights_24khz.pth
- https://github.com/descriptinc/descript-audio-codec/releases/download/0.0.5/weights_16khz.pth
- https://github.com/descriptinc/descript-audio-codec/releases/download/1.0.0/weights_44khz_16kbps.pth
模块
服务方
安装
执行以下命令
git clone https://git.kaiyuancloud.cn/yumoqing/dia16b
cd dia16b
python3 -m venv py3
source py3/bin/activate
pip install git+https://github.com/nari-labs/dia
pip install git+https://git.kaiyuancloud.cn/yumoqing/apppublic
pip install git+https://git.kaiyuancloud.cn/yumoqing/sqlor
pip install git+https://git.kaiyuancloud.cn/yumoqing/ahserver
下载依赖的模型
配置conf/config.json文件
主要修改下载模型存放目录
"dia_model_path" : "/share/models/nari-labs/dia-1___6b",
"dac_model_path": "/share/models/dac/weights.pth",