Welo Data logo

Hydrus - Voice Contributor - English (US)

Welo Data

Intern
0-2 years experience
On-site

Key Skills

Voice Acting
Acting
Theater Performance
Native-Level English Fluency
Conversational Speech Delivery
Script Reading
Spontaneous Speech
Self-Direction
Taking Live Direction
Audio Recording
Microphone Setup
Audio Hygiene
Noise Detection
Clipping Detection
Distortion Detection
Recording Space Setup

Job Description

About The Project Project Hydrus is a scripted and conversational audio data collection project. We're looking for native-level speakers who can deliver natural, everyday-sounding speech - recordings that hold up whether you're reading a script or speaking spontaneously. Please note: the recordings you produce may be used to create synthesized, AI-generated voices for conversational AI products. Full details on how your data is collected, used, shared, and retained are set out in a Consent Form, which you'll review and sign before any paid session recording begins. What You'll Do If Selected Two supervised sessions of approximately 4 hours each, guided in real time by a Session Director on a separate recording platform. Record scripted material in a natural, conversational tone - faithfully, without a polished, performed, or “announcer” delivery. Record unscripted, spontaneous speech on a given topic, matching the same warmth and pace as your scripted takes. Set up and manage your own recording space and equipment for session recording. Deliver all recordings entirely in the required language (with specific dialect), without code-switching into other languages or dialects. Required Experience as a voice actor, actor, or theater performer, or a comparable performance background. Native or native-equivalent fluency in the required language (with specific dialect). Access to a quiet recording space, free of echo and background noise (acoustically treated for session recording, if selected). Owns or will acquire a dedicated, professional-grade microphone for the audition (e.g. RODE NT-USB Mini, Blue Yeti), upgrading to a qualifying USB condenser or XLR mic + audio interface for session recording if selected. Built-in laptop mics, AirPods, earbuds, and headsets are not accepted at any stage. Owns closed-back monitoring headphones (required for session recording if selected). Comfortable both self-directing for the audition and taking live direction from a Session Director in guided sessions. Preferred (nice To Have, Not Required) Professional voice-over experience (dubbing/ADR, audiobook, podcasting) in a conversational rather than commercial/announcer register. Baseline audio hygiene: comfortable setting mic distance and catching noise, clipping, or distortion yourself. An existing home recording setup already treated for sound. Your Data & Privacy Before any session recording, you'll be asked to review and sign a Consent Form covering how your voice recordings are collected, used (including to train synthesized/AI voice models), shared with approved partners, retained, and how to exercise your data rights (access, correction, deletion, and more, depending on where you live). If you'd like to review these terms before applying, contact us and we'll share a copy in advance.

Core Responsibilities

Record scripted and spontaneous conversational speech in English (US) during two supervised sessions of approximately four hours each. Set up and manage a suitable recording space and equipment, and deliver recordings in the required language and dialect without code-switching.

Requirements

Candidates must have voice acting, acting, theater, or comparable performance experience and native or native-equivalent fluency in the required language and dialect. They must have a quiet recording space, a dedicated professional-grade microphone for the audition, closed-back monitoring headphones for sessions, and the ability to self-direct and follow live direction.

About Welo Data

Industry: Technology, Information and Internet

Company size: 1,001-5,000 employees

Welo Data is where frontier AI programs get their data and their edge. We provide the human expertise, training data and evaluation that AI needs to perform in the real world: in physical environments, in multi-step agent workflows and in every language. Physical AI and robotics: end-to-end data programs, from secure lab setup and safety compliance to egocentric video, VLA and UMI data collection, multilingual motion and voice capture, annotation and structured delivery. Multilingual RL environments and agentic evaluation: runnable environments for long-horizon agent work, including tasks, harness, verifiers and expert-defined scoring. Each is authored for the target language and jurisdiction, not translated from English. Model evaluation: custom benchmarks, human-in-the-loop evaluation, RLHF and preference data, red teaming, and relevance and intent assessment. Training data at scale: speech and voice AI, multimodal data, and data collection, generation and annotation across 155+ locales. Behind every program: 500,000+ domain-matched experts, NIMO workforce integrity monitoring in every session, our Inkky data production platform, 14+ secure facilities, 7 ISO certifications, SOC 2 and GDPR. We’re a managed partner, not a self-serve marketplace. Every program is scoped, staffed and monitored by our team. Leading AI teams trust Welo Data, including NVIDIA, Google, Amazon, Workday, Spotify, Squarespace and Dropbox. Follow us for AI research and insights, global job opportunities and stories from our community.

Added Today