coqui

/

XTTS-v2

like 1.2k

maybe your TTS version is older one try updating it (as of now it is 0.20.2) you may try getting help on Discord community,. https://discord.gg/5eXr5seRrv

I am using git+http installed... it's the freshest than ever.

please clone this space and check locally if works out on you
git clone https://huggingface.co/spaces/coqui/xtts

actuall model loading part is here
https://huggingface.co/spaces/coqui/xtts/blob/main/app.py#L49

model inference here (commented out code has direct mode, space uses streaming mode), diffusion_conditioning is dropped on last release
https://huggingface.co/spaces/coqui/xtts/blob/main/app.py#L251

Please check your code again, I look through all code in git repo, there is no such decoed_iteration used anywhere

I see your point now,
If you mean tts.tts_to_file example , decoder_iterations is not used anymore after TTS 0.20.0 (I think you use documentations for TTS 0.17)
Please check on latest examples on docs https://tts.readthedocs.io/en/latest/

This is the example for XTTS in latest docs

import torch
from TTS.api import TTS

# Get device
device = "cuda" if torch.cuda.is_available() else "cpu"

# List available 🐸TTS models
print(TTS().list_models())

# Init TTS
tts = TTS("tts_models/multilingual/multi-dataset/xtts_v2").to(device)

# Run TTS
# ❗ Since this model is multi-lingual voice cloning model, we must set the target speaker_wav and language
# Text to speech list of amplitude values as output
wav = tts.tts(text="Hello world!", speaker_wav="my/cloning/audio.wav", language="en")
# Text to speech to a file
tts.tts_to_file(text="Hello world!", speaker_wav="my/cloning/audio.wav", language="en", file_path="output.wav")

Hello, the code is from this hf repo main page, seems outdated, since it was xttsv2, kindly update it?

btw, what should to do if wanna train more iterations?

I would suggest you join into community, search / ask for answers .
https://discord.com/invite/5eXr5seRrv