English Voice Recording for Public AI Research Dataset (Accents Needed)

We are collecting English speech recordings for an academic and AI research project focused on improving automatic speech recognition (ASR) systems across different accents and speaking styles. The resulting dataset will be released publicly and used by researchers, universities, companies, and developers worldwide to study and improve speech recognition technologies. We are especially looking for speakers with diverse English accents, and both native and non-native English speakers are welcome to apply. Task Overview You will receive a link to an online recording platform. On the platform, you will see multiple rows. Each row contains a sentence or phrase that you need to read, along with a recording option. For each row: Read the sentence or phrase shown. Record your audio using the recording option on the platform. Move to the next row and repeat. You do not need to download, rename, label, or organize any audio files. Please make sure each recording contains only the sentence or phrase shown in that row. Some sentences may sound unusual, grammatically incorrect, or unnatural. This is intentional. Please read every sentence exactly as it appears, even if it seems wrong.   Task Scope The initial paid test task includes: • 222 audio files total • 632 words total • Around 6 minutes of actual recording time on average You do not need to speak in any specific style. Please speak naturally in your own English accent and normal speaking style. Compensation for the test task is $20, paid after the submitted recordings are reviewed and approved. Larger Dataset Contribution If your test recordings are approved and you would like to continue, we may invite you to contribute to the larger training portion of the dataset. The larger task includes approximately: • 8,200 words • Around 1 hour of actual recording time on average For this larger contribution, compensation is $150.   Recording Requirements Please make sure that your recordings meet the following requirements: • Clear English speech • Quiet recording environment • Clear microphone audio • No background music • No artificial voice modification • No audio processing • No AI generated audio A phone, headset, or microphone is acceptable as long as the audio is clear. Public Dataset and Consent This project is part of a speech recognition research effort. Important: your recordings will become part of a public research dataset. This means the recordings will be publicly released and used by researchers, universities, companies, and developers for future AI and speech research. Your name will not be included in the dataset. Recordings will be associated with anonymous speaker IDs. Participants must sign a consent form before beginning the task. The consent document is currently undergoing institutional review as part of the dataset release process. Participants will be asked to sign an updated version before final publication of the dataset. Any updates resulting from this review process are intended to improve participant protections, clarify data usage, and ensure compliance with research standards. No changes will reduce participant rights or protections.   How to Apply If you are interested, please send us: • Your native language • Country of origin • Age at the time of recording, as a number only. Date of birth is not required. • Age: male or female • The signed consent form attached to this post Thank you very much for your interest and contribution to this research.