Lyélé Speech to Text: Preserving a Tonal Language with AI

Lyélé (also spelled Lyélé or Lyele) is a Gur language spoken primarily in the central and northern regions of Burkina Faso, with some speakers in Mali. It belongs to the Niger-Congo language family and is used by the Lyélé people for daily communication, storytelling, and cultural ceremonies. With an estimated 200,000 native speakers, Lyélé is considered a vulnerable language, making digital tools for transcription and subtitling crucial for its preservation.

Why Accurate Speech-to-Text for Lyélé Matters

For many minority languages, the lack of automatic transcription tools means that audio recordings remain inaccessible to text-based search, analysis, and subtitling. Lyélé is no exception. Traditional methods of transcription are slow and require trained linguists. By providing a reliable speech-to-text service for Lyélé, Speechyou empowers communities to:

• Document oral histories and folklore
• Create educational content in the mother tongue
• Generate subtitles for videos and podcasts
• Improve accessibility for deaf and hard-of-hearing speakers
• Enable researchers to process large amounts of spoken data

Transcription Challenges in Lyélé

Lyélé presents several unique challenges for automatic speech recognition. First, it is a tonal language with three contrastive tones: high, mid, and low. A word like "tá" (with high tone) means "to cut," while "tà" (low tone) means "to sell." The ASR model must distinguish these tonal patterns to avoid errors. Second, Lyélé has advanced tongue root vowel harmony, where vowels in a word must agree in terms of tongue root position. This affects suffixation and verb conjugation, and the model must learn these phonological constraints. Third, as a low-resource language, there is limited publicly available speech data. Speechyou overcomes this by using transfer learning from related Gur languages and data augmentation techniques that simulate tonal variations.

Use Cases for Lyélé Transcription

Podcasts and Radio — Local broadcasters can transcribe Lyélé programs for archives and searchable content. Subtitles — Content creators add SRT or VTT subtitles to Lyélé videos, reaching a wider audience. Education — Teachers create subtitled lessons for literacy and language classes. Research — Linguists upload field recordings and receive draft transcriptions, saving hours of manual work. Accessibility — Deaf community members read subtitles of spoken Lyélé content. Oral History — Elders' stories are transcribed and preserved for future generations.

How Speechyou Helps

Speechyou's Lyélé model is fine-tuned specifically for the language's tonal and harmonic features. It handles both northern and southern dialects, though accuracy is highest for clear, single-speaker audio. The service exports text in plain text, SRT, and VTT formats, making it easy to integrate with video editors and media platforms. With unlimited transcription included in the Solo plan, users can process as much Lyélé audio as they need without worrying about per-minute costs.

Getting Started

To transcribe Lyélé audio, simply upload a file (MP3, WAV, MP4, etc.) to Speechyou and select Lyélé as the source language. The system will process the audio and return a text transcript with timestamps. You can then edit the transcript in the built-in editor or export subtitles directly. Whether you are a teacher, researcher, or community member, Speechyou makes Lyélé speech-to-text accessible and affordable.