The official repository for Audio ALBERT

Related tags

AudioAALBERT
Overview

AALBERT

Here is also the official repository of AALBERT, which is Pytorch lightning reimplementation of the paper, Audio ALBERT: A Lite Bert for Self-Supervised Learning of Audio Representation. The original code is in AlbertNew branch of s3prl repo. In the paper, we proposed Audio ALBERT, which achieves performance comparable with massive pre-trained networks in the downstream tasks while having 91% fewer parameters.

drawingdrawing

Dependencies

  • Python 3.8
  • Computing power (high-end GPU) and memory space (both RAM/GPU's RAM) is extremely important if you'd like to train your own model.
  • Required packages and their use are listed requirements.txt.
  • pip install -r requirements.txt

Pretrain Stage

We use LibriSpeech as our pretraining stage dataset. You can download dataset by this link.

  • Stage 1: modify dataset path to your local dataset path:

    • AALBERT: config path: upstream/aalbert/pretrain_config.yaml
          line 16: datarc:
                  {Your dataset key name}: {your local dataset path}
    • Mockingjay: upstream/mockingjay/pretrain_config.yaml
          line 16: datarc:
                  {Your dataset key name}: {your local dataset path}
  • Stage 2: run pretraining script

    python run_pretrain.py -n aalbert_pretrained -u aalbert

    • -n : experiment_name
    • -u : upstream model: {two option: aalbert / mockingjay}
    • model will save on result folder after finish pretraining stage.

Downstream Stage

Here, we take voxceleb1 speaker classification as our downstream task. You can download dataset from their official website.

After pretraining, We can extract the pretrained model feature on different downstream tasks.

  • Stage 1: modify dataset path to your local dataset path
    • voxceleb1_speaker: config path: downstream/voxceleb1_speaker/train_config.yaml
    line  9: datarc:
    line 10:    file_path: {your dataset folder path}
    line 11:    meta_path: {your label file path}
  • Stage 2: run downstream script
    • voxceleb1_speaker:
      python run_downstream.py \
      -c downstream/voxceleb1_speaker/train_config.yaml \
      -g result/pretrain/{your_pretrained_model_folder}/model_config.yaml  \
      -t result/pretrain/{your_pretrained_model_folder}/pretrained_config.yaml \
      -u aalbert \
      -d voxceleb1_speaker \
      -k result/pretrained/{your pretrained_model_folder}/checkpoints/{checkpoint_you_want_to_use.ckpt} \
      -n voxceleb1_result
    • -n: experiment name
    • -c: downstream training config
    • -g: pretrained model config
    • -t: load pretrained model pretrained config
    • -u: upstream model: {two option: aalbert / mockingjay}
    • -d: downstream task name
    • -k: model checkpoint path
    • -f: finetune pretrained model or not, default=False
Owner
pohan
pohan
MIDI-DDSP: Detailed Control of Musical Performance via Hierarchical Modeling

MIDI-DDSP: Detailed Control of Musical Performance via Hierarchical Modeling Demos | Blog Post | Colab Notebook | Paper | MIDI-DDSP is a hierarchical

Magenta 239 Jan 03, 2023
Telegram Bot to play music in VoiceChat with Channel Support and autostarts Radio.

VCPlayerBot Telegram bot to stream videos in telegram voicechat for both groups and channels. Supports live streams, YouTube videos and telegram media

Abdisamad Omar Mohamed 1 Oct 15, 2021
A Python library and tools AUCTUS A6 based radios.

A Python library and tools AUCTUS A6 based radios.

Jonathan Hart 6 Nov 23, 2022
Frescobaldi LilyPond Editor

README for Frescobaldi Homepage: http://www.frescobaldi.org/ Main author: Wilbert Berendsen Frescobaldi is a LilyPond sheet music text editor. It aims

Frescobaldi 600 Dec 29, 2022
A simple music player, powered by Python, utilising various libraries such as Tkinter and Pygame

A simple music player, powered by Python, utilising various libraries such as Tkinter and Pygame

PotentialCoding 2 May 12, 2022
Python I/O for STEM audio files

stempeg = stems + ffmpeg Python package to read and write STEM audio files. Technically, stems are audio containers that combine multiple audio stream

Fabian-Robert Stöter 72 Dec 23, 2022
GNOME powered sound conversion

SoundConverter A simple sound converter application for the GNOME environment. It reads anything the GStreamer library can read, and writes Ogg Vorbis

Gautier Portet 188 Dec 17, 2022
PianoPlayer - Automatic fingering generator for piano scores

PianoPlayer - Automatic fingering generator for piano scores

Marco Musy 571 Jan 02, 2023
A Quick Music Player Made Fully in Python

Quick Music Player Made Fully In Python. Pure Python, cross platform, single function module with no dependencies for playing sounds. Installation & S

1 Dec 24, 2021
python script for getting mp3 files from yaoutube playlist

mp3-from-youtube-playlist python script for getting mp3 files from youtube playlist. Do your non-tech brown relatives ask you for downloading music fr

Shuhan Mirza 7 Oct 19, 2022
:notes: Cross-platform music player

Exaile Exaile is a music player with a simple interface and powerful music management capabilities. Features include automatic fetching of album art,

Exaile 327 Dec 19, 2022
A bot that can play music on Telegram Group and Channel Voice Chats

DaisyXmusic ❤ is the best and only Telegram VC player with playlists, Multi Playback, Channel play and more

TeamOfDaisyX 20 Jun 11, 2021
python wrapper for rubberband

pyrubberband A python wrapper for rubberband. For now, this just provides lightweight wrappers for pitch-shifting and time-stretching. All processing

Brian McFee 106 Nov 28, 2022
ianZiPu is a way to write notation for Guqin (古琴) music.

PyBetween Wrapper for Between - 비트윈을 위한 파이썬 라이브러리 Legal Disclaimer 오직 교육적 목적으로만 사용할수 있으며, 비트윈은 VCNC의 자산입니다. 악의적 공격에 이용할시 처벌 받을수 있습니다. 사용에 따른 책임은 사용자가

Nancy Yi Liang 8 Nov 25, 2022
DCL - An easy to use diacritic library used for diacritic and accent manipulation.

Diacritics Library This library is used for adding, and removing diacritics from strings. Getting started Start by importing the module: import dcl DC

Kreus Amredes 6 Jun 03, 2022
MusicBrainz Picard

MusicBrainz Picard MusicBrainz Picard is a cross-platform (Linux/Mac OS X/Windows) application written in Python and is the official MusicBrainz tagge

MetaBrainz Foundation 3k Dec 31, 2022
Deep learning transformer model that generates unique music sequences.

music-ai Deep learning transformer model that generates unique music sequences. Abstract In 2017, a new state-of-the-art was published for natural lan

xacer 6 Nov 19, 2022
Dataset and baseline code for the VocalSound dataset (ICASSP2022).

VocalSound: A Dataset for Improving Human Vocal Sounds Recognition Introduction Citing Download VocalSound Dataset Details Baseline Experiment Contact

Yuan Gong 58 Jan 03, 2023
Open-Source bot to play songs in your Telegram's Group Voice Chat. Powered by @Akki_ThePro

VcPlayer Telegram Voice-Chat Bot [PyTGCalls] ⇝ Requirements ⇜ Account requirements A Telegram account to use as the music bot, You cannot use regular

Akki ThePro 2 Dec 25, 2021
Python library for handling audio datasets.

AUDIOMATE Audiomate is a library for easy access to audio datasets. It provides the datastructures for accessing/loading different datasets in a gener

Matthias 121 Nov 27, 2022