Native Speaker Voice Recording — en-IN, en-GB, en-FR, en-CN (Short Utterances, Raw WAV)

# Native Speaker Voice Recording — en-IN, en-GB, en-FR, en-CN (Short Utterances, Raw WAV) ## Project Overview We need native/fluent speakers to record short, single-utterance voice samples in four accent/language pairs: **en-IN, en-GB, en-FR, en-CN**. This is a quality-controlled data collection job for speech/audio AI training — every file must meet strict technical specs (details below). Fast turnaround needed: **submission within 24 hours of kickoff.** ## What You'll Do - Record yourself speaking natural, legible sentences or phrases — one utterance per audio file. - Each utterance should be a different, natural sentence (no fixed script — content is flexible as long as it's clear and natural). - Deliver ~15 minutes of total speech per language pair you're recording (roughly **90–150 individual files**, since each file is only 6–10 seconds long). ## Technical Requirements (strict — files outside spec will be rejected) - **1 speaker per file, 1 utterance per file** — not one continuous recording - **Duration:** 6–10 seconds per file, hard limit (nothing shorter, nothing longer) - **Active speech ratio:** at least 55% of each file must be speech — start talking promptly, stop soon after finishing, no long silences at the start/end - **Audio quality:** SNR ≥ 25 dB, no clipping, recorded in a quiet indoor room only - **No post-processing:** no editing, filters, or noise reduction — raw recordings only - **Format:** WAV, 16 kHz, 16-bit, mono - **Consistency:** one recording session per speaker, same device and same room throughout - **Metadata:** must be provided per file, per the spec doc (we'll share this with you directly) ## Who We're Looking For - Native or highly fluent speakers of one or more of: **Indian English (en-IN), British English (en-GB), French (en-FR), Chinese (en-CN — Mandarin-accented English or Mandarin, to confirm with client)** - Access to a quiet, echo-free indoor space - A decent recording device/microphone (phone is fine if quality holds up — no built-in noise reduction/auto-filters enabled) - Comfortable following a precise technical spec and double-checking each file before submission ## Deliverables - 90–150 individual WAV files per language pair (6–10 sec each, 16kHz/16-bit/mono) - Metadata file per the shared spec - All files from a single, consistent recording session (same speaker, device, room) ## Timeline - **Urgent** — we're aiming to submit within 24 hours of hire. Please only apply if you can start immediately and turn this around same-day. ## To Apply, Please Include - Which language pair(s) you can record for (en-IN / en-GB / en-FR / en-CN) - Confirmation you have access to a quiet room and a working microphone/recording device - A short voice sample or link to prior recording work if available - Your availability in the next 24 hours --- *Full QC spec and metadata format will be shared directly with the hired freelancer(s) before recording begins.*