Voice Sample Review

Native Speaker Voice Sample Reviewers Needed We are looking for native speakers to review short voice samples in a target language. These samples will help us choose the best human voices to later create synthetic voices for AI agents that may handle live phone conversations. The goal is to select voices that sound natural, human, conversational, and authentic, not overly polished, theatrical, robotic, or like a voice actor performing a role. Reviewers will listen to voice samples and evaluate them based on: Conversational delivery (1 - 5) Authenticity (1 - 5) Tone and emotional range (1 - 5) Pacing (1 - 5) Audio quality (simple mic and proximity effect check) We are especially looking for reviewers who can evaluate whether a voice of a voice actor sound's authentic and human. Small human imperfections, such as natural pauses, “um” or “uh” moments, slight stumbles, or brief restarts, can be positive when they feel genuine. However, samples should be rejected if they sound overly performed, fake-casual, robotic, theatrical, or unsuitable for real live conversations. Most samples will not be suitable. We only want reviewers to recommend a second round sample that we improve based on feedback, a rejected sample, or a longer recording when the voice is truly strong and would work well as a natural AI agent voice in real-world calls. You must be a native speaker in one of the following languages: Turkish Polish Croatian Hungarian Indonesian Arabic (Lebanese) French (Canada) Korean Russian Dutch Malay Romanian Thai Czech Finnish Norwegian Danish Please apply with a message answering the following: 1. Your native language 2. Your country/region - indicating accent 3. Any experience reviewing audio, voice talent, localization, transcription, QA, linguistics, or language work, and what specifically in relation to this task. 4. A brief note on what makes a voice sound natural and conversational in your language and how you plan on evaluating, we don't want fluff or an AI message, we wan't a person who will simply evaluate and give feedback on how these human sample's sound. This is a short review task per sample (1-2 min), with many sample's in the target language and accent's, with potential for more work as we continue evaluating voices across languages or require more sample's until we find the ideal voice.