add an openai compatible api

[lewangdev]

No description provided.

[lewangdev]

#42

[syq163]

Great job! I'm excited to set up the environment, but I encountered an issue with Python 3.8. Could you please confirm if PR is compatible with Python 3.8, as it's the recommended version to run with EmotiVoice?

[lewangdev]

OK，I will test it on Python 3.8.

[lewangdev]

Hi,
I configured my development environment using the Docker image syq163/emoti-voice:latest, which has Python version 3.10.12.

I used FastAPI to implement the openai compatible API for emoti-voice. And Python 3.10+ has the best and simplest syntax for type hints.

I will change to Python 3.8+ type hints to run with EmotiVoice.

[syq163]

Hi, I configured my development environment using the Docker image syq163/emoti-voice:latest, which has Python version 3.10.12.

I used FastAPI to implement the openai compatible API for emoti-voice. And Python 3.10+ has the best and simplest syntax for type hints.

I will change to Python 3.8+ type hints to run with EmotiVoice.

Thank you for the information. Yeah, I agree 3.10+ might be better.

[syq163]

I have tried Python 3.10 and it works fine. I have two suggestions:

1. I have modified the 'openaiapi.py' file to make it compatible with Python 3.8. Please check it.
   openaiapi.txt

2. I see that you have modified the 'frontend_cn.py' file to make it more robust. However, I suggest that we keep 'frontend_cn.py' unchanged as errors may help us improve it.
   One more thing, it seems that the 'log-config.yml' file is not that necessary, is that right?

[lewangdev]

Great!

it's amazing to see that you have already finished the changes so quickly. I'm sorry that I didn't have time today to modify and support Python 3.8.

log-config.yml is unnecessary and can be deleted.

I agree that you keep 'frontend_cn.py' unchanged.

[syq163]

Thank you for confirming that. Could you please make the necessary updates as we discussed, so that I can easily merge the PR?

Great!

it's amazing to see that you have already finished the changes so quickly. I'm sorry that I didn't have time today to modify and support Python 3.8.

log-config.yml is unnecessary and can be deleted.

I agree that you keep 'frontend_cn.py' unchanged.

[lewangdev]

Thank you for confirming that. Could you please make the necessary updates as we discussed, so that I can easily merge the PR?

Great!
it's amazing to see that you have already finished the changes so quickly. I'm sorry that I didn't have time today to modify and support Python 3.8.
log-config.yml is unnecessary and can be deleted.
I agree that you keep 'frontend_cn.py' unchanged.

ok.

[lewangdev]

Hi, the PR has been updated.

[syq163]

Thank you and all the best!

[lucasjinreal]

Hi, this is good, but if there is no-stream support, we can not say supported openai API now.

At least 2 things not done:

1. Chinese support;
2. streaming tts support;

[newbei]

Can give a demo?

[qqedp045]

curl -X POST -H "Content-Type: application/json" -d '{"voice":"8051","input":"这里有一个改进的版本，使用了一个简单的轮询机制来等待文件生成。"}' http://192.168.21.133:8000/v1/audio/speech > o.mp3
@newbei
input: str
voice: str = '8051'
prompt: Optional[str] = ''
language: Optional[str] = 'zh_us'
model: Optional[str] = 'emoti-voice'
response_format: Optional[str] = 'mp3'
speed: Optional[float] = 1.0

还有其他参数，我最想知道的是prompt的用处。

[qqedp045]

def openai_audio(text: str, outPath: str):
headers = {
'Content-Type':'application/json'
}
url = "http://XXXXXXXX:8080/v1/audio/speech"
query = {
"model":"emoti-voice",
"input":text,
"voice":"984",
"response_format":"mp3",
"speed":1,
}
response = requests.post(url=url, data=json.dumps(query), headers=headers)
# 保存文件
f = open(outPath, "wb")
f.write(response.content)
f.close()

[xushilundao]

this is a test API cpde and it works,you can copy and paste it:
import requests
import json

def openai_audio(text: str, outPath: str):
headers = {
'Content-Type': 'application/json'
}
url = "http://localhost:8000/v1/audio/speech"
query = {
"model": "emoti-voice",
"input": text, # 使用传入的文本参数
"voice": "984",
"response_format": "mp3",
"speed": 1,
}
response = requests.post(url=url, data=json.dumps(query), headers=headers)

# 检查请求是否成功
if response.status_code == 200:
    # 保存文件
    with open(outPath, "wb") as f:
        f.write(response.content)
    print(f"Audio saved to {outPath}")
else:
    print(f"Failed to get audio. Status code: {response.status_code}")
    print(f"Response: {response.text}")

主程序调用部分

if name == "main":
text = "调试成功，调用TTS成功"
output_path = "output.mp3"
openai_audio(text, output_path)

[xushilundao]

_ No description provided. _
I've already run this API code correctly ,but how can I integrate with this to the config file of 火火兔 ?
Cause there is NO key in the api file. How can I config the TTS section in docker-compose.yml?