# Audiogen Codec (agc)
We are announcing the open source release of Audiogen Codec (agc) 🎉. A low compression 48khz stereo neural audio codec for general audio, optimizing for audio fidelity 🎵.
It comes in two flavors:
- agc-continuous 🔄
KL regularized, 32 channels, 100hz.
- agc-discrete 🔢
24 stages of residual vector quantization, 50hz.
AGC (Audiogen Codec) is a convolutional autoencoder based on the DAC architecture, which holds SOTA 🏆. We found that training with EMA and adding a perceptual loss term with CLAP features improved performance. These codecs, being low compression, outperform Meta's EnCodec and DAC on general audio as validated from internal blind ELO games 🎲.
We trained (relatively) very low compression codecs in the pursuit of solving a core issue regarding general music and audio generation, low acoustic quality and audible artifacts, which hinder industry use for these models 🚫🎶. Our hope is to encourage researchers to build hierarchical generative audio models that can efficiently use high sequence length representations without sacrificing semantic abilities 🧠.
This codec will power Audiogen's upcoming models. Stay tuned! 🚀
# Installation
```sh
pip install audiogen-agc
```
# Usage
```python
from agc import AGC
agc = AGC.from_pretrained("Audiogen/agc-continuous") # or "agc-discrete"
audio = torch.randn(1, 2, 480000) # 48khz stereo
z = agc.encode(audio) # (1, 32, 6000) or (1, 24, 3000)
reconstructed_audio = agc.decode(z) # (1, 2, 480000)
```
Raw data
{
"_id": null,
"home_page": "https://github.com/AudiogenAI/agc",
"name": "audiogen-agc",
"maintainer": "",
"docs_url": null,
"requires_python": ">=3.9",
"maintainer_email": "",
"keywords": "",
"author": "Elio Pascarelli",
"author_email": "elio@audiogen.co",
"download_url": "https://files.pythonhosted.org/packages/cb/e6/2b020ca7d3c4092a14a0e8b2d1fc4380bed6cff957ca9b699ee63e9f2c40/audiogen-agc-0.0.1.tar.gz",
"platform": null,
"description": "# Audiogen Codec (agc)\n\nWe are announcing the open source release of Audiogen Codec (agc) \ud83c\udf89. A low compression 48khz stereo neural audio codec for general audio, optimizing for audio fidelity \ud83c\udfb5.\n\nIt comes in two flavors:\n\n- agc-continuous \ud83d\udd04\nKL regularized, 32 channels, 100hz.\n- agc-discrete \ud83d\udd22\n24 stages of residual vector quantization, 50hz.\n\nAGC (Audiogen Codec) is a convolutional autoencoder based on the DAC architecture, which holds SOTA \ud83c\udfc6. We found that training with EMA and adding a perceptual loss term with CLAP features improved performance. These codecs, being low compression, outperform Meta's EnCodec and DAC on general audio as validated from internal blind ELO games \ud83c\udfb2.\n\nWe trained (relatively) very low compression codecs in the pursuit of solving a core issue regarding general music and audio generation, low acoustic quality and audible artifacts, which hinder industry use for these models \ud83d\udeab\ud83c\udfb6. Our hope is to encourage researchers to build hierarchical generative audio models that can efficiently use high sequence length representations without sacrificing semantic abilities \ud83e\udde0.\n\n\nThis codec will power Audiogen's upcoming models. Stay tuned! \ud83d\ude80\n\n\n\n# Installation\n```sh\npip install audiogen-agc\n```\n\n# Usage\n```python\nfrom agc import AGC\n\nagc = AGC.from_pretrained(\"Audiogen/agc-continuous\") # or \"agc-discrete\"\n\naudio = torch.randn(1, 2, 480000) # 48khz stereo\n\nz = agc.encode(audio) # (1, 32, 6000) or (1, 24, 3000)\n\nreconstructed_audio = agc.decode(z) # (1, 2, 480000)\n```\n",
"bugtrack_url": null,
"license": "",
"summary": "Audiogen Codec",
"version": "0.0.1",
"project_urls": {
"Homepage": "https://github.com/AudiogenAI/agc"
},
"split_keywords": [],
"urls": [
{
"comment_text": "",
"digests": {
"blake2b_256": "16e42e4e22f7184f1f2a559c04688476ba776ffbdee3187eb42bd9b9a7261f04",
"md5": "a194c1e4960ca49da3b364d6ba562ec6",
"sha256": "eaa8caf079382f14b45179eb7266dc14cd0a66874522416b3ca0ce5d2fe64e63"
},
"downloads": -1,
"filename": "audiogen_agc-0.0.1-py3-none-any.whl",
"has_sig": false,
"md5_digest": "a194c1e4960ca49da3b364d6ba562ec6",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": ">=3.9",
"size": 7542,
"upload_time": "2024-02-18T21:52:29",
"upload_time_iso_8601": "2024-02-18T21:52:29.176931Z",
"url": "https://files.pythonhosted.org/packages/16/e4/2e4e22f7184f1f2a559c04688476ba776ffbdee3187eb42bd9b9a7261f04/audiogen_agc-0.0.1-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": "",
"digests": {
"blake2b_256": "cbe62b020ca7d3c4092a14a0e8b2d1fc4380bed6cff957ca9b699ee63e9f2c40",
"md5": "a8b82d3ab194895532b4b92fa81da153",
"sha256": "a7289351d853fba80b4f303a4c1abc331af9b2e97382c423879d7ee248ce5586"
},
"downloads": -1,
"filename": "audiogen-agc-0.0.1.tar.gz",
"has_sig": false,
"md5_digest": "a8b82d3ab194895532b4b92fa81da153",
"packagetype": "sdist",
"python_version": "source",
"requires_python": ">=3.9",
"size": 6502,
"upload_time": "2024-02-18T21:52:30",
"upload_time_iso_8601": "2024-02-18T21:52:30.776671Z",
"url": "https://files.pythonhosted.org/packages/cb/e6/2b020ca7d3c4092a14a0e8b2d1fc4380bed6cff957ca9b699ee63e9f2c40/audiogen-agc-0.0.1.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2024-02-18 21:52:30",
"github": true,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"github_user": "AudiogenAI",
"github_project": "agc",
"travis_ci": false,
"coveralls": false,
"github_actions": true,
"lcname": "audiogen-agc"
}
JSON
