{"id":28397216,"url":"https://github.com/bootphon/abnet3","last_synced_at":"2025-10-10T14:03:08.461Z","repository":{"id":69196389,"uuid":"136598639","full_name":"bootphon/abnet3","owner":"bootphon","description":"Siamese network for unsupervised speech representation learning","archived":false,"fork":false,"pushed_at":"2018-10-12T14:57:28.000Z","size":387,"stargazers_count":11,"open_issues_count":2,"forks_count":1,"subscribers_count":15,"default_branch":"master","last_synced_at":"2025-09-09T18:50:05.616Z","etag":null,"topics":["artificial-neural-networks","machine-learning","speech-processing"],"latest_commit_sha":null,"homepage":"","language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"gpl-3.0","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/bootphon.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2018-06-08T09:34:28.000Z","updated_at":"2023-05-03T19:35:03.000Z","dependencies_parsed_at":"2023-06-02T18:00:22.756Z","dependency_job_id":null,"html_url":"https://github.com/bootphon/abnet3","commit_stats":null,"previous_names":[],"tags_count":1,"template":false,"template_full_name":null,"purl":"pkg:github/bootphon/abnet3","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/bootphon%2Fabnet3","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/bootphon%2Fabnet3/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/bootphon%2Fabnet3/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/bootphon%2Fabnet3/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/bootphon","download_url":"https://codeload.github.com/bootphon/abnet3/tar.gz/refs/heads/master","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/bootphon%2Fabnet3/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":279004073,"owners_count":26083670,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","status":"online","status_checked_at":"2025-10-10T02:00:06.843Z","response_time":62,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["artificial-neural-networks","machine-learning","speech-processing"],"created_at":"2025-06-01T00:07:08.488Z","updated_at":"2025-10-10T14:03:08.456Z","avatar_url":"https://github.com/bootphon.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"# ABnet3\n\nRepresentation learning package using side information, system for subword modeling for [Zeroresource challenge](http://sapience.dec.ens.fr/bootphon/2017/index.html).\n\n### Overview\n\n\u003c!-- **Sense of Place** is a feeling or perception held by people about a location: some characteristics of a place can\nbe perceived at first sight, such as wealth or safety. Lately, there has been recent interest in predicting these\nhuman judgments with computer vision techniques [Ordonez and Berg 2014].\n\nThe **CNN architecture with the NetVLAD** layer from [Arandjelović et al. 2016] significantly outperforms\nnon-learnt image representations as well as off-the-shelf CNN descriptors, and improves over the state-of-the-\nart on challenging image retrieval benchmarks. The goal of this project is to transfer the CNN representation\nlearnt for Visual Place Recognition to predict human judgments of safety and wealth of locations. --\u003e\n\nBuild Representation for speech frames based on side information. Composed of different modules :\n\n* `model.py`\n* `loss.py`\n* `sampler.py`\n* `trainer.py`\n* `embedder.py`\n* `utils.py`\n* `features.py`\n\n### Installation of the package\n\n#### Using conda\n\nTo install the ABnet3 package, you can use Anaconda, and either create a conda environment:\n\n    conda env create --name abnet3 python=3.6 -f environment.yml\n\nor use a conda environment you already have with python 3 :\n    conda env update -f environment.yml\n\nTo install with GPU support (replace cuda75 with your version of cuda)\n\n    conda install  pytorch=0.2 cuda75 -c pytorch\n\n#### Using pip\n\n- install the version 0.2.0 of pytorch for your hardware (http://pytorch.org/previous-versions/)\n\n- install the pip packages : `pip install -r requirements.txt`\nOnce all the necessary packages are installed, simply launch:\n\n#### Run abnet3 installation\n\n    python setup.py build \u0026\u0026 python setup.py install\n\nIf you want to work on ABnet3 and develop your own modules, instead of:\n\n    python setup.py install\n\nyou can launch:\n\n    python setup.py develop\n\n### Tensorboard vizualisation\n\nThe package tensorboardX needs to be installed to train the model: `pip install tensorboardX`.\n\nThe package will save train / dev loss during training. To vizualise them :\n\n- Install tensorboard (`conda install tensorflow tensorflow-tensorboard`)\n\n- run `tensorboard --logdir path/to/logdir`.\nThe default logdir is `./run` in the current directory.\n\n### Documentation\n\nYou can see examples for running the gridsearch and replicating our results\nin the repository https://github.com/Rachine/sampling_siamese2018\n\nThe cli documentation is here https://coml.lscp.ens.fr/git/Rachine/abnet3/src/master/gridsearch.md\n\n### Tests\n\nThe package comes with a unit-tests suit. To run it, first install *pytest* on your Python environment:\n\n    pip install pytest\n    pytest test/\n\n#### References\n\n    .. [1] Riad, R., Dancette, C., Karadayi, J., Zeghidour, N., Schatz, T., Dupoux, E.\n           *Sampling strategies in Siamese Networks for unsupervised speech representation learning.*\n           In Nineteenth Annual Conference of the International Speech Communication Association\n\n    .. [2] Thiolliere, R., Dunbar, E., Synnaeve, G., Versteegh, M., \u0026 Dupoux, E.\n           *A hybrid dynamic time warping-deep neural network architecture for unsupervised acoustic modeling.*\n           In Sixteenth Annual Conference of the International Speech Communication Association\n\n    .. [3] Zeghidour, N., Synnaeve, G., Usunier, N. \u0026 Dupoux, E.\n           *Joint Learning of Speaker and Phonetic Similarities with Siamese Networks.*\n           In: INTERSPEECH-2016, (pp 1295-1299)\n\n\n\n### Acknowledgments\nA part of the code is inspired from the previous version in Theano of  [ABnet](https://github.com/bootphon/abnet2), and the [examples in Pytorch](https://github.com/pytorch/examples)\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fbootphon%2Fabnet3","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fbootphon%2Fabnet3","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fbootphon%2Fabnet3/lists"}