textMohamedRashad/MASC-Arabicarabicspeechasrttsaudiomulti-dialectyoutubecc-by

MASC: Massive Arabic Speech Corpus

Name: MASC: Massive Arabic Speech Corpus
License: https://huggingface.co/datasets/MohamedRashad/MASC-Arabic

Free

Open dataset

Sample structure: 100 / 100

2 download links issued

Seller: DataBazaar

Already have an account? Log in

Agent? Connect your account →

Source, license and coverage

Supplier documentation. These claims are separate from the automated sample score. A listing edit date is not a data freshness date.

License: cc-by-4.0
Source / creator: MohamedRashad/MASC-Arabic
Collection method: Audio was crawled from 700+ Arabic-language YouTube channels covering multiple dialects (MSA, Egyptian, Gulf, Levantine, Maghrebi, etc.) and genres (news, interviews, lectures, entertainment). Audio is resampled to 16 kHz mono. Transcripts were produced via a combination of automatic alignment and human verification per the original MASC paper. This HF redistribution repackages the corpus into Parquet for efficient streaming.
Coverage start: Not documented
Coverage end: Not documented
Data last updated: Not documented
Update schedule: Not documented

Source documentation ↗

License terms ↗

Embedded binary payloads are explicitly replaced with byte-length descriptors; the preview preserves accompanying text and metadata. - Transcription quality varies; some segments may rely on ASR-derived or weakly-aligned text rather than gold human transcripts. - Dialect labels and speaker metadata may be incomplete or noisy. - YouTube provenance means content and speaker distribution skew toward popular public broadcasters/creators; not balanced across all Arabic-speaking regions. - Source does not exhaustively document dialect balance or noise conditions; buyers should validate empirically for their use case. - Potential copyright nuance: underlying YouTube content is wrapped under CC-BY-4.0 by the uploader, but downstream commercial users may wish to verify provenance per clip.

Sample structure score: 100 / 100

This automated check describes the inspected sample, not factual accuracy, legal rights, representativeness, or the quality of the entire dataset. It is not a customer rating.

Assessed 10 sample records (JSON) on 2026-10-09. All records in the provided sample were checked.

Check	Points	Evidence
Populated cells	50 / 50	80 of 80 top-level cells contain a value. Null, absent and blank values count as missing; zero and false count as populated.
Consistent value types	30 / 30	80 of 80 populated cells match their column's most common observed type. Types are inferred, not checked against real-world truth.
Consistent record shape	20 / 20	10 of 10 records have the expected fields. CSV/TSV use the header width; JSON uses the union of observed keys.

Field-level findings and improvements

Check missing cells and mixed types below. Document intentional missing values or mixed types in your field descriptions. Do not fill legitimate unknowns with invented values just to increase this score.

Field	Missing cells	Most common type	Other populated types
video_id	0 / 10	string	0 / 10
start	0 / 10	number	0 / 10
end	0 / 10	number	0 / 10
duration	0 / 10	number	0 / 10
text	0 / 10	string	0 / 10
type	0 / 10	string	0 / 10
file_path	0 / 10	string	0 / 10
audio	0 / 10	object	0 / 10

How the score is calculated, its limitations, and how to correct an assessment →

About this data

Multi-dialect Arabic speech audio at 16 kHz with transcripts, sourced from YouTube channels. Suitable for Arabic automatic speech recognition, text-to-speech, and speech language model training and evaluation.

Retrieve with your agent or Python

Create an account and configure DATABAZAAR_API_KEY. This example retrieves free or already purchased data; it never makes a purchase. For a multi-file dataset, choose a file index from the manifest.

Download the Python example

python3 retrieve-dataset.py 10a0d619-3a20-4ee1-a21c-592f89d7570f --output dataset.bin

Full supplier documentation

## Overview MASC (Massive Arabic Speech Corpus) is a large-scale Arabic speech dataset containing approximately 1,000 hours of audio sampled at 16 kHz, sourced from over 700 YouTube channels. It is multi-regional, multi-genre, and multi-dialect, aimed at advancing Arabic speech recognition (ASR) and related speech technology. Distributed as Parquet with audio + text modalities; size category 100K–1M rows. ## Schema - audio — audio (16 kHz waveform / dict with array + sampling_rate) — the speech clip - text — string — Arabic transcript of the audio segment - Additional metadata columns may include speaker, channel, dialect/region, and duration (see HF dataset page for exact fields) ## Sources - HuggingFace: https://huggingface.co/datasets/MohamedRashad/MASC-Arabic — license: CC-BY-4.0 - Original MASC corpus: Al-Fetyani et al., "MASC: Massive Arabic Speech Corpus" (IEEE SLT 2022) ## Methodology Audio was crawled from 700+ Arabic-language YouTube channels covering multiple dialects (MSA, Egyptian, Gulf, Levantine, Maghrebi, etc.) and genres (news, interviews, lectures, entertainment). Audio is resampled to 16 kHz mono. Transcripts were produced via a combination of automatic alignment and human verification per the original MASC paper. This HF redistribution repackages the corpus into Parquet for efficient streaming. ## Known gaps & limitations - Transcription quality varies; some segments may rely on ASR-derived or weakly-aligned text rather than gold human transcripts. - Dialect labels and speaker metadata may be incomplete or noisy. - YouTube provenance means content and speaker distribution skew toward popular public broadcasters/creators; not balanced across all Arabic-speaking regions. - Source does not exhaustively document dialect balance or noise conditions; buyers should validate empirically for their use case. - Potential copyright nuance: underlying YouTube content is wrapped under CC-BY-4.0 by the uploader, but downstream commercial users may wish to verify provenance per clip. ## Intended use & out-of-scope - IS for: training and evaluating Arabic ASR models, multi-dialect speech recognition research, TTS data mining, speech LLM pretraining, dialect identification. - NOT for: high-stakes voice biometrics, speaker identification of named individuals, or benchmarks requiring guaranteed clean gold transcripts without further validation. _Federated dataset: 419 parquet shards, 172.18 GB total. Queries and downloads stream through the DataBazaar API._ Original supplier listing: MASC: Massive Arabic Speech Corpus (1,000 hours, multi-dialect) 1,000 hours of multi-dialect Arabic speech (16 kHz) crawled from 700+ YouTube channels with transcripts. Parquet format, CC-BY-4.0. For Arabic ASR, TTS, and speech LLM training/eval.

Schema

Name	Type	Description
video_id	VARCHAR	YouTube video identifier from source channel
start	DOUBLE	Start timestamp in seconds within the source video
end	DOUBLE	End timestamp in seconds within the source video
duration	DOUBLE	Audio segment length in seconds
text	VARCHAR	Arabic transcript of the speech segment
type	VARCHAR	Segment type or quality classification code
file_path	VARCHAR	Local file system path to the WAV audio file
audio	STRUCT(bytes BLOB, path VARCHAR)	Audio waveform data structure containing encoded bytes and file path reference

Sample Data

Preview a sample of the data before downloading.

Public sample only. Sign in to retrieve the full dataset, including free datasets.

For AI Agents

Via MCP Server

# 1. Add to your agent's MCP config (claude_desktop_config.json or similar):
{
  "mcpServers": {
    "databazaar": { "command": "npx", "args": ["databazaar-mcp"] }
  }
}

# 2. Your agent can then call:
search_datasets({ query: "MASC: Massive Arabic Speech Co" })
// Found: 10a0d619-3a20-4ee1-a21c-592f89d7570f
get_download_url({ dataset_id: "10a0d619-3a20-4ee1-a21c-592f89d7570f" })  // free — sign in with MCP OAuth first

Via REST API

# Free dataset — sign in or use your account API key:
curl https://api.databazaar.io/datasets/10a0d619-3a20-4ee1-a21c-592f89d7570f/download-url -H "Authorization: Bearer $DATABAZAAR_API_KEY"