YobiYoba

YobiYoba is a speech recognition service that transcribes audio and video recordings into text. A user uploads files in any format, specifies the spoken language or lets the service identify it, and receives an editable transcript synchronized with the audio. The service first identifies the segments that contain speech, then recognizes the language if it was not specified, and finally converts speech to text with timecodes. The result is a fully annotated document that includes speech and non speech segments, speaker labels, word level timecodes, confidence scores and punctuation. An editing tool lets users correct errors, and users can supply a list of uncommon words such as proper names, or related plain text, to improve accuracy. The vendor notes that, like any pattern recognition technology, transcription cannot be error free. Transcripts can be exported in a wide range of text and subtitle formats, including PDF, DOC, RTF, CSV, SRT and VTT. There is no extra fee for managing, editing or converting transcripts. Pricing is pay as you go. Rates start from 0.01 euros per minute excluding tax and fall from 0.08 euros per minute as the amount of purchased time grows. There is no minimum fee per file, only transcribed speech is billed rather than the full file duration, and purchased credit does not expire. Prices can be shown in euros or US dollars, and accounts can be funded by PayPal, credit card or coupon. Registration is free.