GigaSpeech 2

by SpeechColab

30,000h multilingual evolution of GigaSpeech — Thai, Indonesian, Vietnamese launch.

TL;DR

30,000h multilingual evolution of GigaSpeech — Thai, Indonesian, Vietnamese launch.

Best for low-resource Southeast Asian language ASR pretraining at scale. Pricing: research-only.

Category
Open source
License
—
Stars
—
Last push
—
Pricing
research-only
Platforms
HuggingFace

What it is

GigaSpeech 2 brings the same crawl-and-pseudo-label pipeline to Thai (10kh), Indonesian (6kh), Vietnamese (6kh). Designed for low-resource ASR pretraining where MLS doesn't reach.

Best for: Low-resource Southeast Asian language ASR pretraining at scale.
Watch out for: Same restrictions as GigaSpeech v1 · YouTube-sourced · noisy-pseudo-labels via cross-lingual transfer + filtering. Cite: Yang et al., Interspeech 2024.

Install / use

from datasets import load_dataset; ds = load_dataset('speechcolab/gigaspeech2', 'th')

Features

Speaker diarizationNo
Word-level timestampsNo
Streaming / real-timeNo
Languages supported3
HIPAA eligibleNo

GigaSpeech 2 vs Whipscribe

FeatureGigaSpeech 2Whipscribe
CategoryOpen sourceTranscription APIs
Pricingresearch-onlyfree beta
Speaker diarizationNoYes
Word timestampsNoYes
StreamingNoNo
Languages399
PlatformsHuggingFaceWeb, API, MCP

Alternatives to GigaSpeech 2

OpenAI Whisper
OpenAI

The reference open-source multilingual ASR model from OpenAI.

OSS · MIT ★ 98.1k
whisper.cpp
Georgi Gerganov

C/C++ port of Whisper — runs on anything, from a Raspberry Pi to Apple Silicon.

OSS · MIT ★ 48.8k
faster-whisper
SYSTRAN

4× faster than reference Whisper using CTranslate2 — production sweet spot.

OSS · MIT ★ 22.3k

Whipscribe is a managed faster-whisper + whisperX service. If you want transcripts without running infrastructure, paste a URL or drop a file in the form below — you'll have a transcript in seconds.