Native Arabic Voice Duo — Scripted Two-Person Dialogue, Studio Quality (Ongoing)

We are building a large scripted speech dataset and need pairs of native Russian speakers who can record two-person dialogue at studio quality, consistently, over an extended project. This is directed voice acting, not narration. Please read the full spec before applying. This role requires TWO speakers You must apply as a pair, or be able to bring a second native Russian speaker you already work with. We record scripted two-person dialogue — both voices reading a shared script expressively and responding to each other in natural conversational rhythm. Requirements for the pair: Both speakers native Russian, each in a quiet or acoustically treated space Both sign the same AI-training release Preferably different gender and/or age range, for speaker balance Each speaker captured to a separate isolated track — not both on a single mic Solo applicants who plan to voice both sides of the conversation will not be considered. The work You record scripted dialogue. Each line is recorded at least twice, and each take carries its own direction note (for example: take 1 "calm, matter-of-fact", take 2 "impatient, rushed"). Delivery must audibly change between takes to match the direction. If your experience is flat read-through narration only, this project will not be a fit. We supply the script and all direction notes. Audio specifications (non-negotiable) 48 kHz, 24-bit WAV Separate isolated track per speaker Treated or acoustically quiet rooms — no room echo, hum, HVAC, traffic, or handling noise Raw and unprocessed: no compression, EQ, noise reduction, de-essing, or normalisation Same microphones, same rooms, same distance and gain across every session No clipping; consistent levels across all files Deliverables per batch Audio files, named to the convention we provide (one file per take) Speaker metadata: gender, age range, accent/region, for both speakers Signed consent and release form per speaker No transcription or annotation work is required on your side. We provide the script and direction notes. Just clean audio, correctly named, plus the release forms. Licensing — please read Recordings will be licensed for AI and machine-learning training use. Both speakers must sign a perpetual, transferable release granting these rights. If either speaker is unwilling to sign, please do not apply — we cannot make exceptions on this point. Volume and process We need 100+ hours in Russian, delivered in ongoing batches. We begin with a small paid test batch of approximately 3 hours of delivered audio. If quality and turnaround are good, we move to continuous batch work. The budget shown is for the initial test batch only. Ongoing batches are quoted and paid separately, per delivered audio hour. Pricing is per delivered audio hour — QA-passed audio in the final deliverable, excluding silence, false starts, and rejected takes. Quote a single rate covering both speakers. To apply Answer the screening questions and attach the requested sample. Applications without a sample will not be reviewed.