Spanish Audio Transcription Reviewer

About OpenTrain

OpenTrain is the #1 platform for finding and building careers in AI training and data labeling. We connect freelancers with specialized, recurring projects that help them grow a lasting freelance practice in the human side of AI.

For contributors this means a single place to build a profile, discover projects, and showcase the AI training work that will make models safer, more accurate, and more conversationally natural.

About AI training work

AI training (also called data labeling or annotation) is the human effort that helps models learn from real examples. In audio and speech projects, people transcribe, evaluate, and validate system outputs so models improve at understanding natural, overlapping, and noisy speech.

This work is ideal for remote, flexible schedules and for people who want entry into a fast-growing tech field: many projects require only language fluency and attention to detail, while specialist tasks pay more for domain expertise.

• 100% remote and flexible—fit tasks around other commitments.
• Accessible: many projects value language skills and careful listening over formal credentials.

The role

OpenTrain is hiring freelance Spanish audio transcription reviewers to evaluate AI-generated transcriptions and diarization for multilingual, multi-speaker conversations. You will validate transcript accuracy, speaker attribution, and timestamp precision against strict acceptance thresholds and WER targets.

This is a contract, part-time role requiring 20+ hours per week. Work is open to worldwide applicants who are fluent or native Spanish speakers (LatAm or Spain). Compensation and project-specific details are provided during the application process.

• Data type: multi-channel AUDIO with multi-speaker conversations.
• Labeling tasks: TRANSCRIPTION and EVALUATION_RATING.
• Employment: Contractor, Part-time (20+ hours/week).

What you'll do

• Review AI-generated transcriptions for accuracy against strict WER (word error rate) targets.
• Validate diarization and speaker attribution across channels and turns.
• Confirm turn-level and word-level timestamps for precision and alignment.
• Check that unnormalized speech (overlaps, interruptions, false starts, disfluencies) is preserved when required.
• Read and validate JSON-formatted outputs to confirm metadata tagging and alignment logic.
• Identify and document transcription, segmentation, and alignment errors in multi-speaker audio.

Requirements

• Fluent or native Spanish proficiency (Latin America or Spain).
• Experience reviewing transcription quality, diarization, segmentation, or ASR outputs.
• Ability to evaluate and judge turn-level and word-level timestamps.
• Strong ear for audio fidelity issues including background noise, channel bleed, and clipping.
• Meticulous attention to detail for speaker attribution, alignment, and low error-rate standards.
• Experience level: Intermediate; comfortable working independently on contract tasks.

Helpful background

• Linguistics or phonetics experience with spontaneous speech.
• Professional transcription or speech data QA experience.
• Previous work with unnormalized conversational speech (overlaps, disfluencies).

How it works / Apply

Create or use your OpenTrain account to apply for this contract role. Your application will be reviewed for language fluency and relevant QA experience; project-specific qualification steps and instructions will be provided to successful applicants.

OpenTrain helps you build a profile of completed projects so you can find more audio and speech work over time. If you meet the requirements and enjoy careful listening and transcript quality evaluation, this role is a direct way to help shape how speech-enabled AI understands real conversations.

• Worldwide applicants welcome; everything is remote.
• Contract details and project timelines are shared during the application and onboarding process.