Hi,

very cool work. The example demos sound very impressive e.g. Elon's voice. I have followed examples in ref [1] and was able to run things with no issues. I then attempted to experiment with adding a few reference voices e.g. David Attenborough and Morgan Freeman based on samples found here [2] and here [3] (>30s high quality audio recordings of their voices). My sample code setup below follows ref [1] and generates outputs with no issues. However, the output generated does not correspond with references for either voice. Are there additional things to consider like length of the reference recording, base speaker tts and other params? Or is it something rather silly?

Sample code:
reference_speaker = 'OpenVoice/resources/morgan_freeman_example.mp3'
target_se, audio_name = se_extractor.get_se(reference_speaker, tone_color_converter, target_dir='processed', vad=True)
save_path = f'{output_dir}/output_morgan_freeman.wav'

Run the base speaker tts

text = "Hello, I am Morgan Freeman, and you are inside the Matrix."
src_path = f'{output_dir}/tmp.wav'
base_speaker_tts.tts(text, src_path, speaker='default', language='English', speed=1.0)

Run the tone color converter

encode_message = "@MyShell"
tone_color_converter.convert(
audio_src_path=src_path,
src_se=source_se,
tgt_se=target_se,
output_path=save_path,
message=encode_message)

References:
[1] https://github.com/myshell-ai/OpenVoice/blob/main/demo_part1.ipynb
[2] https://en.wikipedia.org/wiki/File:Sir_David_Attenborough_BBC_Radio4_Desert_Island_Discs_29_Jan_2012_b01b8yy0.flac
[3] https://en.wikipedia.org/wiki/File:Morgan_freeman_bbc_radio4_film_programme_12_09_2008_b00dbcdn.flac