bigvgan_v2_44khz_128band_512x

[wishgale]

bigvgan has this version. When can you replace it?

Thanks.

[yaozengwei]

We could train this version and release it later.

[yaozengwei]

@wishgale We have uploaded this version, mel_44k_128band_512x. Please see README for details. Could try:

from flow2gan import get_model
from flow2gan.models.modules import LogMelSpectrogram

step = 4  # Could set step to 1,2,4
# 24kHz model
# model_name = "mel_24k_base"
# hf_model_name = f"libritts-mel-{step}-step"
# hf_model_name = f"universal-24k-mel-{step}-step"  # For universal audio model

# 44kHz model
model_name = "mel_44k_128band_512x_base"
hf_model_name = f"universal-44k-mel-128band-512x-{step}-step"  # For universal audio model

# Required model will be downloaded from HuggingFace Hub automatically
model, model_cfg = get_model(model_name=model_name, hf_model_name=hf_model_name)

[wishgale]

Thanks, that is great!