Brazilian Portuguese Speech Recordings From Mobile Phones

Aggregating 1,044 hours of Brazilian Portuguese speech recorded by 2,038 native speakers using mobile phones. The recording text was designed by linguistic experts and covers general interactive, in-car, and home categories, with texts manually proofread for high accuracy.

Use Cases

• Train an automatic speech recognition (ASR) model on 1,044 hours of Brazilian Portuguese speech data recorded via mobile phones.
• Fine-tune a speech model for specific domains using the linguist-designed text categories covering general interactive, in-car, and home scenarios.
• Analyze speaker variation and acoustic characteristics across a pool of 2,038 Brazilian native speakers.

Strengths

• Large volume of 1,044 hours of speech data.
• Recordings from 2,038 distinct Brazilian native speakers, providing speaker diversity.
• Recording text designed by linguistic experts and manually proofread for high accuracy.
• Recordings captured using mainstream Android phones and iPhones, reflecting real-world mobile audio conditions.

Limitations

• The dataset is described as a sample of a larger paid dataset, so the full data is not available.
• Specific audio file formats, sample rates, and metadata columns are not detailed in the provided input.
• The temporal coverage and specific geographic distribution within Brazil are not specified.

Provenance

Source

Nexdata

Collection Method

Recorded by 2,038 Brazilian native speakers using mainstream Android phones and iPhones, with text designed by linguistic experts.

Freshness

Last updated on 2025-04-28.

Geography

Brazil