Case Study: Benevis speeds up speech-to-text conversion with Sloboda Studio

Benevis speeds up speech-to-text conversion by 50% with Sloboda Studio

Benevis, a Norwegian company providing an innovative speech-to-text conversion service, faced the challenge of automating the transcription of Norwegian audio. Their goal was to process raw speech-to-text results to split them into well-structured sentences with correct punctuation, supporting both formal and informal language variants. They partnered with Sloboda Studio to develop a machine learning-based post-processing mechanism.

Sloboda Studio implemented a solution using Google's BERT neural network for natural language processing. Their team refined the network's rules and used an extensive text training kit to fine-tune the algorithm for the Norwegian language. As a result, Benevis received an automated feature that halves the speech-to-text conversion time, meaning the process now takes only half the length of the audio clip itself.

Benevis