How does the YobiYoba transcription process work?

Updated

YobiYoba is a speech recognition service that converts audio and video recordings into time-coded, editable transcripts. The process has four steps: upload your files, specify the language and process them, verify and edit the transcripts, and download them in your chosen format. You need to register for an account before you can start, and pricing is pay-as-you-go from 0.01 € per minute (excl. tax), billed only for transcribed speech rather than total file duration.

The four-step process

Step 1: Upload your file(s)

Upload audio or video files in any format from any source. There is no stated restriction on file type or origin, so you can bring recordings from different devices or platforms without converting them first.

Step 2: Specify the language and process the file(s)

You tell the service which language is spoken. You can also provide textual information about your data to help the automatic processing. If the language is not known or not specified in advance, YobiYoba identifies the language being spoken on its own.

Step 3: Verify and edit the transcripts

The transcript is synchronized with the audio, so you can listen and read at the same time while checking the output. You can modify the transcripts directly. This editing step matters because, as YobiYoba states, speech transcription cannot be error free — like any pattern recognition technology, it produces results that need review.

Step 4: Download transcripts

Choose from a wide selection of transcript formats, including SRT, XML, RTF, PDF, DOC, CSV, and VTT.

What the output contains

The processing produces a fully annotated document rather than plain text. According to YobiYoba, the result includes:

  • Speech and non-speech segments
  • Speaker labels
  • Words with time codes
  • High quality confidence scores
  • Punctuation marks

Internally, the result is an XML document that is converted on demand into the text and subtitling formats listed above.

How the speech-to-text conversion works

YobiYoba describes three underlying steps:

  1. The software identifies the audio segments containing speech.
  2. It recognizes the language being spoken if it is not known a priori.
  3. It converts the speech segments to time-coded text.

Pricing and billing conditions

Item Detail
Pricing model Simple pay-as-you-go
Rate range From 0.08 € to 0.01 € per minute (excl. tax), degressive with purchased time
Minimum fee per file None
What you pay for Amount of transcribed speech, not file duration
Extra fees None to manage, edit, or convert transcripts
Credit expiry Transcription credit never expires
Payment methods PayPal, credit card, or coupon
Tax note All prices include EU VAT

Practical points to keep in mind

  • Register first. The site repeatedly offers "Register for Free," which is the entry point to the service.
  • Editing is part of the workflow, not optional cleanup. Budget time for the verify-and-edit step, especially for recordings with background noise, multiple speakers, or specialized vocabulary.
  • You are billed for speech, not silence. Long files with quiet stretches cost less than their duration might suggest.
  • Format choice happens at the end. Since conversion to PDF, SRT, VTT, and others is on-demand and free of extra fees, you can decide the output format after reviewing the transcript.