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

Expressive text-to-speech (TTS) aims to synthesize speeches with human-liketones, moods, or even artistic attributes. Recent advancements in expressiveTTS empower users with the ability to directly control synthesis style throughnatural language prompts. However, these methods often require excessivetraining with a significant amount of style-annotated data, which can bechallenging to acquire. Moreover, they may have limited adaptability due tofixed style annotations. In this work, we present FreeStyleTTS (FS-TTS), acontrollable expressive TTS model with minimal human annotations. Our approachutilizes a large language model (LLM) to transform expressive TTS into a styleretrieval task. The LLM selects the best-matching style references fromannotated utterances based on external style prompts, which can be raw inputtext or natural language style descriptions. The selected reference guides theTTS pipeline to synthesize speeches with the intended style. This innovativeapproach provides flexible, versatile, and precise style control with minimalhuman workload. Experiments on a Mandarin storytelling corpus demonstrateFS-TTS's proficiency in leveraging LLM's semantic inference ability to retrievedesired styles from either input text or user-defined descriptions. Thisresults in synthetic speeches that are closely aligned with the specifiedstyles.

Expressive TTS Driven by Natural Language Prompts Using Few Human Annotations