Japanese–English Bilingual Conversational Data Collection

Job ID: 40065029

Budget: $30 – $250 USD

We are launching a large-scale Japanese–English bilingual conversational data collection project for AI and speech technology training.
We are accepting both individual participant applications and a project manager application to oversee recruitment and operations.



Option 1: Individual Participants (Japanese–English Speakers)

Role

Participate in natural, unscripted two-person conversations recorded for AI training purposes.

Requirements
• Native Japanese residents, aged 18–60
• Fluent spoken English (slight accent acceptable)
• Natural speaking style (no reading or acting)
• Ability to follow technical and consent requirements
• Each participant may take part once only

Recording Conditions
• Quiet indoor environment (≤50 dB, no echo)
• Smartphone + microphone recording
• Multi-track recording (each speaker on a separate channel)
• WAV format, mono, 48kHz / 16–32bit

Session Structure
• Topic-based free conversation
• 5–30 minutes per topic
• Maximum 2 hours per speaker pair
• Natural turn-taking, overlaps, and interruptions allowed



Option 2: Project Manager / Team Lead (Required)

Role

We are seeking one experienced coordinator to manage the end-to-end execution of the project.

Responsibilities
• Recruit and onboard qualified Japanese–English speakers
• Organize speaker pairs and recording schedules
• Ensure compliance with recording and quality standards
• Manage participant metadata and consent documentation
• Coordinate file naming, validation, and final delivery
• Act as the main point of contact throughout the project

Required Skills
• Experience in data collection, linguistic projects, or crowdsourcing
• Strong organizational and people-management skills
• Familiarity with audio recording standards
• Ability to enforce quality control and compliance rules
• Fluency in English (Japanese is a strong advantage)



Quality & Compliance (Applies to All Roles)
• Audio must be free of background noise, echo, and third-party voices
• Reading-style or scripted speech is not accepted
• All participants must sign electronic consent forms
• Voice consent must be recorded prior to sessions
• Inconsistent or unverifiable submissions will be rejected



Deliverables
• Clean, validated multi-track audio files
• Speaker metadata (age, gender, region)
• Signed consent agreements and voice consent files
• Fully structured and annotated dataset



Use Cases
• Speech recognition (ASR)
• Conversational AI and voice assistants
• Bilingual language model training
• Accent-robust NLP systems