Bam Asr Early: Bambara Speech Audio for Automatic Speech Recognition

RobotsMali's Bam Asr Early dataset is a collection of Bambara language audio for automatic speech recognition (ASR). It primarily combines the Jeli-ASR dataset and the Mali-Pense data curated by Aboubacar Ouattara, with an additional hour of audio featuring children's voices reading books. The dataset was last updated on March 18, 2025.

Use Cases

• Training automatic speech recognition models based on Bambara audio recordings.
• Fine-tuning ASR models for children's speech based on the described audio of children reading.
• Researching multilingual or low-resource speech recognition based on the combined sources of Bambara data.

Strengths

• Includes at least one hour of recently collected audio featuring children's voices, which suggests diversity in speaker demographics.
• Combines data from multiple described sources (Jeli-ASR, Mali-Pense, RobotsMali lab), suggesting a composite collection.

Limitations

• Row count, column names, and file formats are unknown, which limits suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

RobotsMali, combining Jeli-ASR dataset, Mali-Pense data from Aboubacar Ouattara, and RobotsMali AI4D Lab recordings.

Collection Method

Likely a compilation of existing datasets and newly collected audio.

Freshness

Last updated 2025-03-18 18:08:30; freshness should be verified.

Geography

Likely Mali or West Africa, given the focus on the Bambara language.