GigaSpeech

by SpeechColab (consortium)

10,000h English ASR corpus — audiobook + podcast + YouTube blend, multiple subsets.

TL;DR

10,000h English ASR corpus — audiobook + podcast + YouTube blend, multiple subsets.

Best for large-scale English ASR with mixed-domain coverage (read + spontaneous + spoken). Pricing: research-only.

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

What it is

GigaSpeech is a 10,000-hour English ASR corpus spanning audiobooks (LibriVox), podcasts, and YouTube. Five subset sizes (XS→XL) for staged training. Apache-2.0 metadata + source-license-bound audio. Request form gated.

Best for: Large-scale English ASR with mixed-domain coverage (read + spontaneous + spoken).
Watch out for: Apache-2.0 metadata · UNDERLYING AUDIO has source-license restrictions (YouTube TOS, podcast copyright) · request-form access · subsets XS(10h)/S(250h)/M(1000h)/L(2500h)/XL(10000h). Cite: Chen et al., Interspeech 2021.

Install / use

from datasets import load_dataset; ds = load_dataset('speechcolab/gigaspeech', 'xs')  # request access

Features

Speaker diarizationNo
Word-level timestampsNo
Streaming / real-timeNo
Languages supported1
HIPAA eligibleNo

GigaSpeech vs Whipscribe

FeatureGigaSpeechWhipscribe
CategoryOpen sourceTranscription APIs
Pricingresearch-onlyfree beta
Speaker diarizationNoYes
Word timestampsNoYes
StreamingNoNo
Languages199
PlatformsHuggingFace, GitHubWeb, API, MCP

Alternatives to GigaSpeech

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.