Output just gibberish
Hello there!
I've added the model to E2/F5 (Pinokio) via
hf://marduk-ra/F5-TTS-German/f5_tts_german_1010000.safetensors
hf://marduk-ra/F5-TTS-German/vocab.txt
and it produces output - but it's just gibberish. no german.
What do I need in the "Config" field?
Am I missing something?
Thanks!
it's a problem with the newer versions of F5-TTS itself. i reverted to 0.6.2 and it works fine again.
Ah, I see. How would I downgrade within Pinokio or install another version in parallel? prost
i never used pinokio, i install everything in python venvs. just went to the release page on github, downloaded the source code .zip of version 0.6.2 and replaced the files
Ah, I see. How would I downgrade within Pinokio or install another version in parallel? prost
maybe my answer is to late, but:
the change f5-tts did with new version is: old 15 second input audio, new 12 second input audio
if you change your audio files for voice cloning to 12 seconds, you can use them im new f5 tts versions with this german model.
(if using gradio app, change the model settings in the gradio to old setting you
pro tipp: with new f5 tts use a blank txt file now, as the audio transcription work inside the gradio app now)