What Information Is Included in YobiYoba Transcripts?

Updated

YobiYoba returns a fully annotated transcript rather than plain text. According to the site, the processing result is an XML document containing speech and non-speech segments, speaker labels, words with time codes, confidence scores, and punctuation marks. You can then convert that document on demand into text or subtitle formats such as PDF, DOC, RTF, CSV, SRT, or VTT.

What the transcript contains

Element What it tells you
Speech and non-speech segments Which parts of the recording contain speech and which do not
Speaker labels Which speaker produced each segment
Words with time codes When each word occurs in the audio
Confidence scores How reliable the recognition is for the transcribed content
Punctuation marks Sentence-level punctuation in the output

This structure matters because it lets you verify the transcript against the audio and edit it, rather than accepting a single block of unverified text.

How the transcript is produced

YobiYoba describes a three-step conversion process:

  1. The software identifies the audio segments containing speech.
  2. It recognizes the spoken language if it was not known and specified in advance.
  3. It converts the speech segments into time-coded text.

The result is the annotated XML document described above. The site notes that the service can also accept textual information about your data to help the automatic processing.

How you get your output

The workflow on the site is presented as four steps:

  1. Upload audio or video files in any format from any source.
  2. Specify the language and process the files.
  3. Verify and edit the transcripts, synchronized with the audio.
  4. Download transcripts in your chosen format.

The format list includes SRT, XML, RTF, PDF, DOC, CSV, and VTT.

What to keep in mind

YobiYoba states that speech transcription, like any pattern recognition technology, cannot be error free, which is why an editing step is provided. Treat the confidence scores and the audio-synchronized editor as the tools for catching those errors before you export.

Pricing is pay-as-you-go, starting from 0.01 € (excl. tax) per minute, with rates degressive from 0.08 € to 0.01 € per minute depending on purchased time. Billing is based on the amount of transcribed speech, not file duration, and transcription credit does not expire. The site lists PayPal, credit card, and coupon as ways to credit your account.