TutlAit v1: a crowdsourced Moroccan Tamazight speech dataset with Arabic transcriptions and regional accent labels
Authors: Mohamed-Amine Chadi, Ezzahra Ait El Arbi, Ismail Khayoub, Aymane Fadili, Yassine Ennhili, Jadjigua Bouali, Hanane Inhid, Mohammed Ameksa, Hajar Mousannif
Organizations: Department of Computer Science, Faculty of Sciences Semlalia, Cadi Ayyad University, Boulevard Prince My Abdellah, B.P. 2390, Marrakech, 40000, Morocco.
Tamazight (Amazigh) is, together with Arabic, one of the two official languages of Morocco, yet it remains severely under-resourced for speech technology: pub licly available labelled audio is scarce, generally lacks information on the regional variety spoken, and is often of uneven transcription quality. This article describes the TutlAit dataset, a corpus of Moroccan Tamazight speech paired with Modern Standard Arabic text and explicit regional accent labels. The data were collected with TutlAit, a purpose-built crowdsourcing web application (React 18 front end, Django 5 / Django REST Framework back-end, PostgreSQL database). Native speakers recruited through targeted LinkedIn and Instagram campaigns created an account, declared their regional variety (Atlas, Souss, Rif or other) and demographic information, and then contributed through two workflows: Text-to Audio, in which an Arabic sentence is displayed and the volunteer records its oral Tamazight rendering in the browser, and Audio-to-Text, in which a Tamazight excerpt is played and the volunteer types its Arabic transcription. A complemen tary set of segments was obtained from freely accessible Tamazight audiovisual media, segmented and annotated with ELAN and imported through a bulk CSV/ZIP pipeline. Every upload is converted server-side to 16kHz mono WAV, hashed with SHA-256 for duplicate rejection, checked for duration bounds and validated by an administrator. The dataset contains 13,384 audio files totalling 75,231 seconds (approximately 20.9 hours, about 3.01GB). The Atlas variety accounts for 9,956 files (14.08h) and the Souss variety for 3,378 files (6.75h); small Rif (22 files) and Kabyle (28 files) subsets are also included. The corpus can be reused for speech recognition, speech translation and accent identification for Moroccan Tamazight.
Figures & tables
Column
Type
Description
annotation_id
integer
Unique identifier of the contribution; matches the WAV file name.
audio_path
string
Relative path of the WAV file inside tutlait_audio.zip .
text
string
Modern Standard Arabic text (source sentence for Text-to-Audio, transcription for Audio-to-Text).
accent
categorical
Regional variety: atlas , souss , rif , kabyle .
mode
categorical
Workflow: text_to_audio , audio_to_text or bulk_import .
duration_s
float
Clip duration in seconds.
Table 1: Columns of the metadata index metadata.csv .
Metric
Value
Number of audio files
13,384
Total duration (seconds)
75,231
Total duration (hours)
≈ 20.9 h
Total size
≈ 3.01 GB
Mean duration per file
5.62 s
Median duration per file
4.03 s
Table 2: Global statistics of the TutlAit dataset.
Duration interval (s)
Number of files
Percentage
(0, 2]
3,153
23.56%
(2, 4]
3,487
26.05%
(4, 6]
2,442
18.25%
(6, 8]
1,590
11.88%
(8, 10]
844
6.31%
(10, 15]
1,049
7.84%
Table 3: Distribution of audio files by duration.
Figure 1: Distribution of audio file durations in the TutlAit dataset.
Regional variety
Number of files
Percentage
Atlas
9,956
74.39%
Souss
3,378
25.24%
Kabyle
28
0.21%
Rif
22
0.16%
Total
13,384
100%
Table 4: Number of audio files per regional variety.
Regional variety
Total duration (hours)
Percentage
Atlas
14.08
67.61%
Souss
6.75
32.42%
Rif
0.05
0.24%
Kabyle
0.02
0.10%
Total
20.90
100%
Table 5: Total recorded duration per regional variety.
Figure 2: Number of audio files per regional variety.
Figure 3: Total recorded duration (hours) per regional variety.
Figure 4: Overall workflow of corpus collection, segmentation and annotation.
Figure 5: General architecture of the TutlAit platform.
We present Cardamom, a micro-dialectal Arabic speech dataset designed to support fine-grained evaluation and adaptation of automatic speech recognition (ASR) systems. Community-curated by native speakers familiar with the represented varieties, Cardamom contains approximately 40 hours of transcribed YouTube speech spanning 21 micro-dialects across Egypt, Jordan, Lebanon, Mauritania, Palestine, and Saudi Arabia. Each segment is annotated with one or more operational micro-dialect labels, code-switching information, and utterance-level perceived gender, enabling analysis of sub-country variation that is obscured by conventional country-level labels. We describe the collection and annotation process, motivate the micro-dialect inventory linguistically, and benchmark four multilingual ASR systems in zero-shot and adapted settings. The strongest zero-shot system obtains 43.47% aggregate WER, with particularly high error rates on Mauritanian and Lebanese varieties; adaptation on Cardamom reduces its WER to 35.21%. Audio-based identification experiments further show that the annotations provide a learnable prediction target, with a dedicated classifier reaching 85.57% accuracy on 21-way micro-dialect identification. Cardamom provides a resource for studying localized dialectal variation and developing Arabic speech systems with broader regional coverage.
Bashar Talafha, Samar M. Magdy, Aisha Alansari +36
The University of British Columbia · King Fahd University of Petroleum and Minerals · Al al-Bayt University +13
Arabic automatic speech recognition (ASR) faces unique challenges due to diglossia, extensive regional dialect variation, and limited speech resources. Existing speech datasets often focus on single dialects or large-scale broadcast/web data, leading to trade-offs between linguistic diversity and annotation quality. We present BULBUL, a multi-dialect Arabic ASR dataset collected from 275 speakers in 11 Arab countries. BULBUL includes structured dialect and sub-dialect coverage, as well as recordings of classical Arabic and modern standard Arabic spoken by participants in their native dialectal accents to support accent-aware modeling. The quality of the recordings was ensured through a two-level human verification process. We further benchmark a range of recent ASR systems, establishing strong baselines for modern dialectal and accented Arabic ASR.
We present the Thiomi Dataset, a large-scale multimodal corpus spanning ten African languages across four language families: Swahili, Kikuyu, Kamba, Kimeru, Luo, Maasai, Kipsigis, Somali (East Africa); Wolof (West Africa); and Fulani (West/Central Africa). The dataset contains over 601,000 approved sentence-level text annotations and over 385,000 audio recordings, collected through a dedicated community data collection platform involving over 100 contributors. To validate the dataset's utility, we train and evaluate ASR, MT, and TTS models, establishing baselines across all languages. Our best ASR system achieves 3.24% WER on Swahili (Common Voice), reducing prior academic SOTA from 8.3% to 3.24% (5.1 percentage point absolute, 61% relative reduction), and 4.3% WER on Somali. The dataset will be published on HuggingFace. We describe the collection platform, quality assurance workflows, and baseline experiments, and discuss implications for African language technology infrastructure.