Edresson Casanova, Ryan Langman, Paarth Neekhara, Shehzeen Hussain, Jason Li, Subhankar Ghosh, Ante Jukić, Sang-gil Lee
{"title":"低帧率语音编解码器:专为快速高质量语音 LLM 训练和推理而设计的编解码器","authors":"Edresson Casanova, Ryan Langman, Paarth Neekhara, Shehzeen Hussain, Jason Li, Subhankar Ghosh, Ante Jukić, Sang-gil Lee","doi":"arxiv-2409.12117","DOIUrl":null,"url":null,"abstract":"Large language models (LLMs) have significantly advanced audio processing\nthrough audio codecs that convert audio into discrete tokens, enabling the\napplication of language modeling techniques to audio data. However, audio\ncodecs often operate at high frame rates, resulting in slow training and\ninference, especially for autoregressive models. To address this challenge, we\npresent the Low Frame-rate Speech Codec (LFSC): a neural audio codec that\nleverages finite scalar quantization and adversarial training with large speech\nlanguage models to achieve high-quality audio compression with a 1.89 kbps\nbitrate and 21.5 frames per second. We demonstrate that our novel codec can\nmake the inference of LLM-based text-to-speech models around three times faster\nwhile improving intelligibility and producing quality comparable to previous\nmodels.","PeriodicalId":501284,"journal":{"name":"arXiv - EE - Audio and Speech Processing","volume":"1 1","pages":""},"PeriodicalIF":0.0000,"publicationDate":"2024-09-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Low Frame-rate Speech Codec: a Codec Designed for Fast High-quality Speech LLM Training and Inference\",\"authors\":\"Edresson Casanova, Ryan Langman, Paarth Neekhara, Shehzeen Hussain, Jason Li, Subhankar Ghosh, Ante Jukić, Sang-gil Lee\",\"doi\":\"arxiv-2409.12117\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Large language models (LLMs) have significantly advanced audio processing\\nthrough audio codecs that convert audio into discrete tokens, enabling the\\napplication of language modeling techniques to audio data. However, audio\\ncodecs often operate at high frame rates, resulting in slow training and\\ninference, especially for autoregressive models. To address this challenge, we\\npresent the Low Frame-rate Speech Codec (LFSC): a neural audio codec that\\nleverages finite scalar quantization and adversarial training with large speech\\nlanguage models to achieve high-quality audio compression with a 1.89 kbps\\nbitrate and 21.5 frames per second. We demonstrate that our novel codec can\\nmake the inference of LLM-based text-to-speech models around three times faster\\nwhile improving intelligibility and producing quality comparable to previous\\nmodels.\",\"PeriodicalId\":501284,\"journal\":{\"name\":\"arXiv - EE - Audio and Speech Processing\",\"volume\":\"1 1\",\"pages\":\"\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2024-09-18\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"arXiv - EE - Audio and Speech Processing\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/arxiv-2409.12117\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"arXiv - EE - Audio and Speech Processing","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/arxiv-2409.12117","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
Low Frame-rate Speech Codec: a Codec Designed for Fast High-quality Speech LLM Training and Inference
Large language models (LLMs) have significantly advanced audio processing
through audio codecs that convert audio into discrete tokens, enabling the
application of language modeling techniques to audio data. However, audio
codecs often operate at high frame rates, resulting in slow training and
inference, especially for autoregressive models. To address this challenge, we
present the Low Frame-rate Speech Codec (LFSC): a neural audio codec that
leverages finite scalar quantization and adversarial training with large speech
language models to achieve high-quality audio compression with a 1.89 kbps
bitrate and 21.5 frames per second. We demonstrate that our novel codec can
make the inference of LLM-based text-to-speech models around three times faster
while improving intelligibility and producing quality comparable to previous
models.