Slow Inference with Python 3.9(For MeloTTS), fast on 3.11

[ban1989ban]

log_message(f"Starting TTS inference for speaker ")
start_tts_time = time.time()
text = "Hello, this is a test sentence for text to speech"
output_path_1 = 'en-india-1.wav'
model.tts_to_file(text, speaker_ids['EN-INDIA'], output_path_1, speed=speed)
log_message(f"TTS inference completed in {time.time() - start_tts_time:.2f} seconds.")

Same code working in python 3.11 in avg 0.20 - 0.5 sec
on python 3.9 : 4-5 sec (since openvoice not working in 3.11)

[dosubot]

There are known compatibility issues with OpenVoice when using Python 3.11, which is why it is recommended to use Python 3.9 to avoid errors like 'ImportError: numpy.core.multiarray failed to import' ^[1]. The performance difference you're experiencing might be due to optimizations or changes in Python 3.11 that are not present in Python 3.9. Unfortunately, specific details about performance optimizations or issues related to different Python versions in the MeloTTS model are not documented in the repository.

To continue talking to Dosu, mention @dosu.

---

^{Help Dosu learn! Give it feedback: Great Response | Irrelevant Answer | Incorrect Sources | Too Verbose | Hallucination | Bug Report | Other}