Folio3 developed a complete audio-to-text transcription service
Folio3 developed a complete audio-to-text transcription service that compared the transcripted text with the targeted text and provided a score based on accuracy.
The customer is a US-based e-learning platform provider that allows users to acquire new language speaking skills by connecting them with certified native speakers while offering premium eLearning products at competitive price points.
The client wanted a complete audio-to-text transcription engine implemented backed with an API that would convert the speech to text and cross-check the output with a target text to provide an overall score for accuracy.
Folio3 utilized its Converse Smartly platform to deliver a complete audio-to-text transcription service coupled with API support to process two different inputs, i.e. audio files and the target text. Folio3 developed an entire web interface to allow users to upload audio files and cross-check that with the target text for that specific lesson.
The solution was powered by the IBM Watson Speech Recognition API to calculate the accuracy of the transcriptions against the target text. Folio3 also helped the client train the IBM Watson speech model to accurately process different accents and pronunciations from different US regions.
With Folio3’s help, the client was able to provide its users with an easy-to-use web interface where they could easily upload their audio recordings for different lessons and check them with the appropriate texts while getting real-time feedback on accuracy.
This led to far smoother user experiences and helped users work on specific parts of their speech with ease.