[BUG🐛] Error when using OpenAI server with speech generation

[elvinzade]

Bug Description

I encountered an error when attempting to use the OpenAI server for speech generation. Below is the error message and the code that caused it. This error occurs when i run script with enhance_speech": False otherwise it works but it provide low quality rather than actual coqui fine tuned model

Minimal Reproducible Example

from openai import OpenAI  
import base64  

text = "Hello"

reference_audio_path = "audio.wav"

client = OpenAI(  
        api_key="your-api-key",  base_url="url")  
with open(reference_audio_path, "rb") as f:  
    audio_data = base64.b64encode(f.read()).decode('utf-8')  # Configure speech parameters  
    speech_params = {  
        "enhance_speech": False, "max_ref_length": 60, "gpt_cond_len": 30, "gpt_cond_chunk_len": 4, "temperature": 0.75, "top_p": 0.85, "top_k": 50, "repetition_penalty": 20.0, "length_penalty": 1.0, "language": "en"}  
  
# Generate speech  
response = client.audio.speech.create(  
 model="xttsv2", voice=[audio_data], input=text, response_format="wav", speed=1.0, extra_body=speech_params)  
  
# Save the audio file  
response.write_to_file("output.wav")

Error Logs

\x03G\xb9\xb3\xd9\x00\xdf\x99\xa4\xc0\x93a{\xc8\x99\\\xf7\xc3\xba\xf2\x83\xc0\xbb\xb9\x92\xe2i\x9b_-\xfc\xc2D27\xe4\x86\xd3\xa3\xd8\xd4\xac\xe7\xe5\xb6\xa8b\xff\xa3Cm:\xce{[\xba\x80\x84S\xd9\xde\xb5\xa6\x0e\x94\x8e\\&\x93j\xec\xe5|\x18\xc4]\xae^\x11e\xf0\xa2%a\xe4\xd8\xb0\x8c\xefG\x10\x00\xd6{\x8aL\xf4\xb2\x16\x05\x99\x80\xbf\x0c\xcb]L\x7fn8\xdcJ\x9bt\xfc\x81\x01\x0e\x97\xa5\x80OggS\x00\x04x\xdd\x01\x00\x00\x00\x00\x00}7\x7fE,\x00\x00\x00\xe4*z\x80\x01TX\x03Nk\xc7(S\xb9\xe4\x8a\xd0A\xfe\xb4\n3\x9fM\xbf\x02\xa2\xf8\xf0\xbe\xd1}\x84\x13(5O\xb8M\xa2\xcc\xa7\xf0\xed\x8e\xa4\xc0)\x90\x0c\x87\x9bw\r\xb4\x90\x15F\xe3\xae\x18\x96,B\x07\x0c\xfao\xf0\x93\xf1m\x889R7\x9c\x80#D\x9e\xf3\xccy\x84m.\xaf\x80\xc0': System error.

[mlinmg]

@elvinzade Can't seem to reproduce on my end, it works fine, have you tried with a fresh env?