Susu Speech-to-Text: Bringing AI Transcription to a West African Language

Introduction

Susu (Sosoxui) is a Mande language spoken by approximately 2.5 million people across Guinea, Sierra Leone, and Guinea-Bissau. It is one of the six national languages of Guinea and holds cultural significance as the language of oral traditions, radio broadcasts, and daily communication. Yet, for years, Susu speakers lacked access to modern speech-to-text tools. Most major transcription services — including Google Speech-to-Text, Amazon Transcribe, and Rev — do not support Susu at all. This digital gap has hindered content creation, education, and accessibility for Susu-speaking communities.

Why Accurate Speech-to-Text for Susu Matters

Accurate transcription of Susu audio is essential for several reasons:

• Media Preservation: Susu radio stations produce hours of news, talk shows, and music daily. Transcribing these broadcasts creates searchable archives for journalists and historians.
• Education: Susu-language educational videos and online courses need subtitles and transcripts to improve literacy and reach learners worldwide.
• Oral History: Griots and elders pass down history through spoken word. Transcribing these recordings safeguards them for future generations.
• Accessibility: Deaf and hard-of-hearing Susu speakers benefit from subtitles on video content.

Technical Challenges of ASR for Susu

Transcribing Susu is not trivial. The language has a two-tone system (high and low) that distinguishes meaning. For example, the word "bà" means "father" while "bá" means "rice." Generic ASR models often miss these tonal differences, leading to errors. Additionally, Susu exhibits vowel harmony based on advanced tongue root (ATR), affecting how suffixes are pronounced. Code-switching with French is common in urban areas, further complicating recognition.

Speechyou's Susu model is trained specifically on these phenomena. Using a transformer-based architecture and a curated dataset of Susu speech from Guinea and Sierra Leone, it achieves high accuracy even on tonal and code-switched input.

Use Cases for Susu Speech-to-Text

• Podcast Transcription: Susu podcasters can generate transcripts and show notes automatically.
• YouTube Subtitling: Upload videos and get SRT or VTT files in Susu (Latin script) instantly.
• Research: Linguists can transcribe field recordings for analysis of Susu phonology and syntax.
• Healthcare: Clinics can transcribe Susu patient consultations for better record-keeping.
• Social Media: Convert voice messages and video captions into text for wider engagement.
• Language Learning: Create interactive Susu lessons with synchronized transcripts.

How Speechyou Delivers

Speechyou offers a fully automated Susu speech-to-text service with a simple interface. Users can upload audio or video files, paste YouTube links, or record directly in the app. The output includes plain text, full transcript, and subtitle files (SRT, VTT, ASS). The model supports both Coastal and Interior Susu dialects, with continuous improvements from user feedback.

Unlike competitors that ignore Susu, Speechyou provides a dedicated solution. The Solo plan includes unlimited transcription, making it affordable for individual creators and small organizations. For teams, the Pro plan offers additional features like speaker diarization and batch processing.

Conclusion

Susu speech-to-text is no longer a missing piece. With Speechyou, speakers of one of West Africa's most important languages can now transcribe, subtitle, and preserve their audio content with AI accuracy. Whether you're a journalist in Conakry, a teacher in Kambia, or a content creator in the diaspora, Speechyou brings the power of speech recognition to Susu.