In a significant step towards preserving African languages, the first African API and transcription platform has been unveiled. This platform focuses on the Igbo language, spoken by over 50 million people globally, primarily in southeastern Nigeria.
Introducing IgboSpeech, an advanced Automatic Speech Recognition (ASR) model designed specifically for Igbo speech and text, boasting a 29 Word Error Rate (WER). The developers aim to improve the model continuously and release it alongside the Igbo API.
The platform, accessible at speech.igboapi.com, allows users to drag and drop audio files, record, upload, or select Igbo audio for transcription into text. It promises accurate transcription, leveraging industry-leading ASR technology to ensure precision.
IgboSpeech’s real-time processing capability enables seamless transcription during live events, video calls, and more. The API employs state-of-the-art machine learning models to handle various accents and speech patterns effectively.
The project’s organizers are calling for passionate volunteers to help with recording and reviewing Igbo audio, as well as contributing ideas for new features to drive the initiative forward.
For more information and to join the effort, visit speech.igboapi.com.