audioku/meta-transfer-learning

Implementation of meta-transfer-learning for ASR and LM (ACL 2020)

41
/ 100
Emerging

This project helps create automated speech recognition (ASR) systems that can accurately transcribe audio where speakers frequently switch between multiple languages in a single conversation, often called code-switching. It takes recordings of mixed-language speech and produces transcribed text. This is designed for AI/ML researchers or engineers who are building robust ASR systems for multilingual environments, especially in low-resource settings.

No commits in the last 6 months.

Use this if you need to develop an ASR system capable of handling code-switched speech efficiently, particularly when you have limited mixed-language data but access to more abundant single-language datasets.

Not ideal if you are looking for a ready-to-use application for transcribing audio or if your speech recognition needs are solely for a single language.

multilingual-speech-recognition natural-language-processing code-switching low-resource-languages speech-to-text
Stale 6m No Package No Dependents
Maintenance 0 / 25
Adoption 8 / 25
Maturity 16 / 25
Community 17 / 25

How are scores calculated?

Stars

52

Forks

11

Language

Python

License

MIT

Last pushed

Jul 30, 2020

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/transformers/audioku/meta-transfer-learning"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.