{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/audio-word2vec-unsupervised-learning-of-audio","title":"Audio Word2Vec: Unsupervised Learning of Audio Segment Representations using Sequence-to-sequence Autoencoder","arxiv_id":"1603.00982","date":"2016-03-03","proceeding":null,"authors":["Yu-An Chung","Chao-Chung Wu","Chia-Hao Shen","Hung-Yi Lee","Lin-shan Lee"],"abstract":"The vector representations of fixed dimensionality for words (in text)\noffered by Word2Vec have been shown to be very useful in many application\nscenarios, in particular due to the semantic information they carry. This paper\nproposes a parallel version, the Audio Word2Vec. It offers the vector\nrepresentations of fixed dimensionality for variable-length audio segments.\nThese vector representations are shown to describe the sequential phonetic\nstructures of the audio segments to a good degree, with very attractive real\nworld applications such as query-by-example Spoken Term Detection (STD). In\nthis STD application, the proposed approach significantly outperformed the\nconventional Dynamic Time Warping (DTW) based approaches at significantly lower\ncomputation requirements. We propose unsupervised learning of Audio Word2Vec\nfrom audio data without human annotation using Sequence-to-sequence Audoencoder\n(SA). SA consists of two RNNs equipped with Long Short-Term Memory (LSTM)\nunits: the first RNN (encoder) maps the input audio sequence into a vector\nrepresentation of fixed dimensionality, and the second RNN (decoder) maps the\nrepresentation back to the input audio sequence. The two RNNs are jointly\ntrained by minimizing the reconstruction error. Denoising Sequence-to-sequence\nAutoencoder (DSA) is furthered proposed offering more robust learning.","url_abs":"http://arxiv.org/abs/1603.00982v4","url_pdf":"http://arxiv.org/pdf/1603.00982v4.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"audio-word2vec-unsupervised-learning-of-audio","repo_url":"https://github.com/earthspecies/audio-embeddings","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok"}}],"tasks":[{"task_slug":"decoder","task_name":"Decoder"},{"task_slug":"denoising","task_name":"Denoising"},{"task_slug":"dynamic-time-warping","task_name":"Dynamic Time Warping"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/1603.00982","atlas_url":"https://app.syntology.ai/?focus=1603.00982","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}