English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Language model based text-to-speech (TTS) models, like VALL-E, have gainedattention for their outstanding in-context learning capability in zero-shotscenarios. Neural speech codec is a critical component of these models, whichcan convert speech into discrete token representations. However, excessivetoken sequences from the codec may negatively affect prediction accuracy andrestrict the progression of Language model based TTS models. To address thisissue, this paper proposes a novel neural speech codec with time-invariantcodes named TiCodec. By encoding and quantizing time-invariant information intoa separate code, TiCodec can reduce the amount of frame-level information thatneeds encoding, effectively decreasing the number of tokens as codes of speech.Furthermore, this paper introduces a time-invariant encoding consistency lossto enhance the consistency of time-invariant code within an utterance and forceit to capture more global information, which can benefit the zero-shot TTStask. Experimental results demonstrate that TiCodec can not only enhance thequality of reconstruction speech with fewer tokens but also increase thesimilarity and naturalness, as well as reduce the word error rate of thesynthesized speech by the TTS model.

Fewer-token Neural Speech Codec with Time-invariant Codes