Preserving Borôro with AI Speech-to-Text

Borôro, known natively as Boe Wadáru, is the ancestral language of the Bororo people in the Brazilian state of Mato Grosso. It belongs to the Macro-Jê language family and is critically endangered, with only about 1,000 speakers remaining. The language is deeply intertwined with the cultural identity of the Bororo, used in rituals, storytelling, and daily communication. However, as younger generations increasingly adopt Portuguese, the transmission of Borôro is at risk.

Why Accurate Speech-to-Text is Crucial for Borôro

Accurate speech-to-text technology can play a key role in language revitalization. By converting spoken Borôro into written text, communities can create archives of oral traditions, produce educational materials, and add subtitles to videos. This not only preserves the language but also makes it more accessible to learners and researchers. Speechyou's Borôro speech-to-text model is designed specifically for this purpose, with high accuracy on the language's unique phonetic features.

Challenges in Transcribing Borôro

Borôro presents several challenges for automatic speech recognition:

• Phonemic vowel length: Short and long vowels are contrastive. For example, 'e' (short) vs 'ee' (long) can change word meaning. Generic ASR models often fail to distinguish these.
• Glottal stops: The glottal stop /ʔ/ is a consonant in Borôro, appearing frequently. It is often omitted by models that haven't been trained on it.
• Nasalized vowels: Vowels can be nasalized, which affects meaning. The model must capture nasal formants.
• Limited data: As a low-resource language, there is little existing transcribed audio. Speechyou uses transfer learning from related Macro-Jê languages and community recordings to build a robust model.

Use Cases for Borôro Transcription

Oral History Preservation

Elders hold vast knowledge of Borôro myths, songs, and historical accounts. Transcribing these recordings ensures that future generations can access them in written form. With Speechyou, a 30-minute recording can be transcribed in minutes, rather than hours of manual work.

Educational Materials

Schools in the Bororo territory can use transcribed texts to create bilingual readers and worksheets. Subtitled videos of cultural practices help students learn both language and traditions.

Community Media

Borôro radio stations and YouTube channels can generate subtitles for their content, reaching a wider audience. This includes live streams of ceremonies and community meetings.

Research and Documentation

Linguists and anthropologists study the Borôro language and culture. Speechyou provides a fast way to get draft transcripts of field recordings, which can then be refined and annotated.

How Speechyou Delivers

Speechyou's Borôro model is available in the app and supports audio and video files up to 2 hours. The transcription is near real-time, and you can export SRT or VTT subtitles. The model handles both Eastern and Western dialects, and works with noisy recordings typical of outdoor events. Unlike other tools that ignore endangered languages, Speechyou includes Borôro in its standard plan with unlimited transcription. This means no per-minute charges, making it accessible for communities with limited budgets.

Getting Started

To transcribe Borôro audio, open the Speechyou app, upload your file, and select "Borôro (Latin)" as the language. The AI will process the audio and produce a transcript. You can then edit the text, add timestamps, and export subtitles. The process is simple and does not require technical expertise. We encourage Borôro speakers to contribute feedback on transcriptions to help improve the model further.

By embracing technology, the Bororo community can ensure that their language thrives in the digital age. Speechyou is proud to support this mission.