German Audio QA Specialist — Transcription & Diarization

About OpenTrain

OpenTrain is the centralized platform where people build careers in AI training and data labeling. We help freelancers find and grow work in a fast-moving industry, build a unified portfolio, and apply to projects quickly.

OpenTrain AI is the hiring and contracting organization for this role. Contributors work remotely and help shape how modern AI systems learn from human examples.

Why AI training matters

AI training (a.k.a. data labeling or annotation) is the human side of building intelligent systems: people prepare, transcribe, and evaluate real-world examples that modern models learn from.

This work is remote and flexible, accessible to many skill levels, and gives contributors a direct role in improving speech and language technology.

The role

You will review multi-channel conversation recordings and the human-validated transcripts and diarization that accompany them. The focus is on evaluation-grade checks: audio fidelity, channel isolation, timestamps, speaker IDs, metadata correctness, and safety/privacy.

This is a contractor, part-time assignment: a 4-week remote contract requiring 20+ hours per week. The dataset consists of transcribed, multi-channel conversations with JSON metadata.

What you'll do

Perform detailed, repeatable QA on audio and annotation outputs and provide clear corrective feedback when a reviewer disagrees with an annotator's work.

• Listen to multi-channel audio to detect fidelity problems like bleed, clipping, low gain, and background noise.
• Verify verbatim transcriptions and check turn-level and word-level timestamp accuracy.
• Confirm diarization quality, speaker attribution, overlapping speech handling, and channel isolation.
• Inspect JSON-based metadata and tags for logical consistency and correct values.
• Audit sessions for PII, toxic, or other sensitive content and flag according to guidelines.
• Assign pass/fail or agree/disagree outcomes and write actionable review comments when corrections are needed.

Requirements

Candidates must meet the core skills below to ensure reliable evaluation of high-quality speech data.

• Native proficiency in German (spoken and written) for transcription and review tasks.
• Strong ear for audio fidelity and subtle recording defects (bleed, clipping, low gain, background noise).
• Experience checking strict verbatim transcription and timestamp precision (turn- and word-level).
• Ability to evaluate complex multi-speaker dialogue including overlapping speech and diarization.
• Comfort reviewing annotation data and metadata tags with care and precision, including JSON logic.
• Careful judgment when identifying PII, toxic, or sensitive content and documenting findings.

Helpful background

These qualifications are not required but will make you especially effective in this role.

• Linguistics or phonetics training.
• Professional transcription or QA experience.
• Prior work with multilingual speech datasets or diarization review.

Project details & how to apply

Assignment type: contractor (part-time). Duration: 4 weeks. Hours: 20+ hours per week. Work location: remote — worldwide applicants accepted.

Task types: transcription verification and evaluation rating for multi-channel, human-validated audio; metadata and safety/privacy checks.

To apply, create or sign in to your OpenTrain account, complete the profile steps, and submit your application. OpenTrain will provide access instructions, project guidelines, and the QA rubric after selection.