Remote Voice AI Quality Assessor

Inworld; Mountain View (CA); Remoto; USD 28.000 - 76.000; Tempo pieno

Descrizione del lavoro

Inworld is a research lab specializing in realtime AI for consumer applications. We seek listeners with a trained ear to evaluate voice samples, judge naturalness, and label attributes like gender and accent.

You may work remotely across the United States and support multilingual evaluations. Responsibilities include ranking prosody, timing, and pronunciation, reporting issues, and helping engineers improve speech models.

Competenze

• Background in audio production, broadcast, voice acting, linguistics, or phonetics.
• Ability to hear subtle differences in prosody, timing, pronunciation.
• Strong, careful judgment across large batches of samples.
• Willingness to troubleshoot access and use beta web tools.
• Excellent written English for reporting findings.
• Ability to work day-to-day with potential weekend availability.

Mansioni

• Listen to speech samples and judge naturalness and expression.
• Score voices using rubrics for clarity and delivery.
• Label attributes like gender, age, accent, and tone.
• Judge clone voice accuracy against source speaker.
• Review batches for audio quality before release.
• Test new features in pre-release environments and report bugs.
• Write clear findings for engineers to act on.
• Evaluate pronunciation and naturalness in native language for multilingual roles.

Conoscenze

Audio production background

Prosody and phonetics listening

Consistent judgment

Written English ability

VPN access familiarity

Spreadsheet skills

Weekend availability

Strumenti

VPN access

Spreadsheets (Excel/Sheets)

Descrizione del lavoro

Inworld is a research lab specializing in realtime AI for consumer applications. We seek listeners with a trained ear to evaluate voice samples, judge naturalness, and label attributes like gender and accent.

You may work remotely across the United States and support multilingual evaluations. Responsibilities include ranking prosody, timing, and pronunciation, reporting issues, and helping engineers improve speech models.