We are looking for experienced ASR & Transcription QA Specialists to support high-fidelity Automated Speech Recognition (ASR) benchmarking and transcript quality evaluation.

The role involves reviewing and correcting machine-generated transcripts, identifying transcription errors, performing speaker diarization, handling overlapping conversations, and ensuring high levels of accuracy and adherence to transcription guidelines.

Key Responsibilities

- Long-Form ASR Review: Review and rectify machine-generated transcripts across diverse audio qualities, background noise profiles, and non-standard acoustics.
- WER Reduction & Error Tagging: Identify, categorize, and correct substitution, insertion, and deletion errors to drive models toward sub-3% Word Error Rates.
- Speaker Diarization & Audio Segmentation: Apply precise speaker labeling, handle rapid conversational turn-taking, and accurately capture overlapping speech and cross-talk.
- Disfluency & Paralinguistic Formatting: Accurately capture false starts, stutters, filler words, vocalized pauses,



and non-verbal acoustic markers based on verbatim protocols.
- Style Guide & Convention Adherence: Switch seamlessly between strict verbatim and clean-read conventions across specialized domain glossaries.
- OCR & Narrative QA Support: Cross-reference extracted text from unstructured document images and video captions to ensure complete semantic and character fidelity.

Candidate Requirements

- 2-3+ years of professional experience in Transcription, Captioning, Subtitling, or related transcription services.
- Strong experience in reviewing and editing audio transcripts.
- Typing speed of 65+ WPM with 99% accuracy.
- Familiarity with Word Error Rate (WER), timecoding, speaker diarization, and transcription style guides.
- Robust English listening comprehension and written communication skills.
- Experience with professional transcription tools will be an added advantage.

📌 ASR & Transcription QA Specialist (Noida)
🏢 Innodata India
📍 Noida