Spaces:
Running
on
Zero
Synthesis audio
When synthesizing audio, she usually starts with the end of the reference I input. I want to ask how to solve this。
example :
synthesis text : Hey! How are you today.
my reference audio : she is walking in the street, this is amazing.
synthesis text in audio : amazing，Hey! How are you today.
This is an example, I synthesize the audio with chinese word.
and IF i use the remove silence sometimes the problom solved.
What is this problem !?
Hi,
How long is your reference audio sample?
Thanks!
<15seconds，6、12、13 seconds
Hi, is this issue only happening in Chinese or also in English?
I only use in Chinese.
Not the above text, i just give some same example in my Chinese synthesis
Hi, is the reference audio complete (for example, does it include the entire phrase as intended), or is it cut off mid-sentence?
Yes it is completed, and one 13s audio is just cut from the 1minutes audio and i cut it in the first 13second with completed sentences.
Could you try with a different reference sample and see if the issue still occurs?
Yes, sometimes happen sometimes not, but if i chose the Remove Silences button the error rate will be low.
Hmm, that's strange. Please open an issue here to ask the author of F5-TTS.
OK! thank you~!