Mirabelle Transcription Project

What is the project about?

DataForce is searching for detail-oriented transcribers and QA reviewers to support a multilingual podcast transcription and annotation project. The role involves reviewing ASR-generated transcripts, correcting errors, segmenting audio, and applying annotation tags to ensure high-quality training data for speech recognition systems. 

Task Description:

As a Transcriber you will: 

• Review and correct ASR-generated transcripts for accuracy and completeness.   
• Adjust audio segmentation and ensure proper speaker separation.   
• Transcribe speech verbatim, including fillers, repetitions, and colloquialisms.   
• Apply annotation tags (e.g., overlapping speech, non-verbal sounds, unintelligible speech).   
• Assign and verify speaker labels across segments.   
• Meet productivity and quality requests.

Who is eligible?

To participate you should meet the following qualifying criteria:

• Idiomatic fluency in one of the following languages: Chinese, Turkish, Swedish, Russian, Romanian, Polish, Norwegian, Dutch, Korean, Japanese, Italian, Indonesian,Hungarian, Hindi, French, Spanish, German, Danish, Arabic (MSA), Czech, Vietnamese, Thai, English (US), Portuguese, Finnish.  
• Advanced English.  
• Access to a computer with reliable internet connection.  
• Strong listening and comprehension skills.  
• Experience in transcription, subtitling, translation or audio annotation preferred.   
• High attention to detail and consistency.   
• Ability to follow detailed guidelines and QA standards.

Where is the project taking place?

This is a remote project taking place at the comfort of your home.

When?

The project will start on 27th of April 2026 and is estimated to run for 2 weeks. Estimated weekly workload is minimum of 15 hours.