Emotional Audio Data Project Voice Actor Needed – Swedish

We’re hiring voice talents to record 60 minutes of scripted audio demonstrating a range of emotions (e.g., excitement, contempt, confusion) and speaking styles (e.g., narrative, whispered, reading to children). Recordings will be used to train text-to-speech (TTS) models. There is no public broadcast or advertising usage, and nothing will be created in your likeness.

Important:
You must submit an audition that follows the technical requirements listed below to be considered for this project.

Project Overview & Usage:

Scope: 1 hour of finished speech (solo; no dialogues). Fully scripted; scripts provided.

Usage: Internal model training/evaluation only. No public-facing usage and no voice cloning.

Script quality: Scripts are machine-translated; please read as written to the best of your ability without letting translation quality compromise a natural, believable delivery.

Artistic Direction:

• Aim for genuine, natural performances that do not feel performed or stiff.
• Maintain clear, confident delivery and correct pronunciation.

Qualifications

Technical / Audio Requirements

• Format: WAV, stereo, 16-bit
• Sample rate: 48 kHz
• Quality: Very clean audio (noise floor ≤ −60 dB), no reverb, no clipping, minimal plosives.
• Processing: Do not apply denoisers, EQ, compression, normalization, or any plugins—raw audio only.

Revisions: Include up to 2 rounds of revisions due to talent error per script.

Additional Information

To Apply: Submit your application in this job post and send your audition with your proposal in the Upwork platform with the guidance of our team. You will be using the attached script in the Upwork job post. Make sure you are recording the language you are applying for.