{"id":44404,"library":"bigvgan","title":"BigVGAN","description":"BigVGAN is a universal neural vocoder with large-scale training. The current version is 2.4.1, with active development and monthly releases.","status":"active","version":"2.4.1","language":"python","source_language":"en","source_url":"https://github.com/NVIDIA/BigVGAN","tags":["vocoder","audio","generation","neural","bigvgan"],"install":[{"cmd":"pip install bigvgan","lang":"bash","label":"Install from PyPI"}],"dependencies":[{"reason":"Core dependency for neural network operations","package":"torch","optional":false},{"reason":"Audio processing and array operations","package":"numpy","optional":false},{"reason":"Signal processing utilities","package":"scipy","optional":true}],"imports":[{"note":"","symbol":"BigVGAN","correct":"from bigvgan import BigVGAN"},{"note":"load_checkpoint is in bigvgan.utils in v2.4+","wrong":"from bigvgan import load_checkpoint","symbol":"load_checkpoint","correct":"from bigvgan.utils import load_checkpoint"}],"quickstart":{"code":"import torch\nfrom bigvgan import BigVGAN\nfrom bigvgan.utils import load_checkpoint\n\nmodel = BigVGAN()\ncheckpoint = 'path/to/checkpoint.pt'\nload_checkpoint(model, checkpoint)\nmodel.eval()\n# Generate audio\nmel = torch.randn(1, 80, 64)  # Example mel-spectrogram\nwith torch.no_grad():\n    audio = model(mel)  # Returns waveform tensor","lang":"python","description":"Load a pretrained BigVGAN model and generate audio from a mel-spectrogram."},"warnings":[{"fix":"Update import statement to `from bigvgan.utils import load_checkpoint`.","message":"In version 2.4.0, the checkpoint loading API changed from `from bigvgan import load_checkpoint` to `from bigvgan.utils import load_checkpoint`.","severity":"breaking","affected_versions":">=2.4.0,<2.5"},{"fix":"Use `BigVGAN()` and `load_checkpoint` instead.","message":"`BigVGAN.from_pretrained` is deprecated in v2.4 and will be removed in v2.5.","severity":"deprecated","affected_versions":">=2.4,<2.5"},{"fix":"Ensure mel-spectrogram has 80 bins or adjust model configuration.","message":"The input mel-spectrogram must have the correct number of bins (80 by default). Using 128 bins from HiFi-GAN will cause shape mismatch errors.","severity":"gotcha","affected_versions":">=2.0"}],"env_vars":null,"search_vec":"'2.4.1':17 'activ':19 'audio':25 'bigvgan':1,2,28 'current':14 'develop':20 'generat':26 'larg':10 'large-scal':9 'month':22 'neural':6,27 'releas':23 'scale':11 'train':12 'univers':5 'version':15 'vocod':7,24","created_at":"2026-06-07T12:49:58.077325+00:00","updated_at":"2026-06-07T12:49:58.077325+00:00","problems":[{"fix":"Reshape or generate mel with 80 bins. Example: `mel = librosa.feature.melspectrogram(y=y, sr=sr, n_mels=80)`.","cause":"Input mel-spectrogram has wrong number of channels (e.g., 128 instead of 80).","error":"RuntimeError: Given groups=1, weight of size [32, 256, 7], expected input[1, 128, 64] to have 256 channels, but got 128 channels instead"},{"fix":"Upgrade to latest version: `pip install --upgrade bigvgan`. For older versions, import from `bigvgan` directly.","cause":"Using an older version of bigvgan (< 2.4.0) where the utilities were in a different module.","error":"ModuleNotFoundError: No module named 'bigvgan.utils'"},{"fix":"Use `model = BigVGAN()` and load checkpoint manually with `load_checkpoint(model, path)`.","cause":"Attempting to use deprecated `from_pretrained` method removed in v2.5.","error":"AttributeError: 'BigVGAN' object has no attribute 'from_pretrained'"}],"ecosystem":"pypi","meta_description":null,"install_score":null,"quickstart_score":null,"quickstart_tag":null,"pypi_latest":null,"cli_name":null,"cli_version":null,"type":"library","homepage":null,"github":null,"docs":null,"changelog":null,"pypi":null,"npm":null,"openapi_spec":null,"status_page":null,"smithery":null,"categories":["ai-ml"],"base_url":null,"auth_type":null,"provenance":{"verified_status":"passing","verified_at":"2026-06-29","last_verified":"2026-06-29","next_check":"2026-07-29","install_tag":null}}