Vibeleaderboard
Index / tool

CavaAlign v1

huggingface.co
Visit huggingface.co
Category
AI Tools
Rank
No. 2754Tools index
Pricing
Open Source
Type
TOOL
Use case
Data Processing
Interfaces
Client SDK
Date

About

CavaAlign v1 is an ONNX model that takes a recording and its text and returns a start and end time for every word, covering 35 languages in a 16 MB file. The model card reports that 88% of word starts on 70 LibriSpeech speakers fall within 50 ms of forced-alignment truth. It includes Python and Kotlin aligners and handles text that starts before the audio, such as chapter titles.

Tags

forced-alignmentonnxspeechword-timestampsaudiobooksmultilingual

Media

CavaAlign v1

Comments (0)

No comments yet

Editorially curated, with community endorsements as a secondary signal. Corrections welcome.