-
Notifications
You must be signed in to change notification settings - Fork 325
Open
Description
This is my code on ghcr.io/speaches-ai/speaches:0.8.2-cuda-12.4.1.
from pathlib import Path
from openai import OpenAI
openai = OpenAI(base_url="http://192.168.1.100:8000/v1", api_key="cant-be-empty")
model_id = "speaches-ai/Kokoro-82M-v1.0-ONNX"
voice_id = "jm_kumo"
res = openai.audio.speech.create(model=model_id,voice=voice_id,input="今日は月曜日で、夜の10時38分です。こんにちは。日本語で話します。とても静かな夜の時間ですね。",response_format="wav",speed=1)
with Path("output.wav").open("wb") as f:
f.write(res.response.read())
When I listen to output.wav it does not sound like Japanese at all. I believe with Kokoro you have to specify which language to use or else it will try to speak it as English. Is there a way to do this?
Metadata
Metadata
Assignees
Labels
No labels