Source Job

• Listen to short audio clips in Russian and evaluate them using a defined rubric.
• Accurately identify target accents from provided audio samples.
• Compare multiple recordings and assess which one sounds more natural in relation to the target accent.

Welo Data, part of Welocalize, is a global AI data company with 500,000+ contributors delivering high-quality, ethical data to train the world’s most advanced AI systems. We’re building smarter, more human AI with a diverse community in 100+ countries.

• Listen to short audio clips in Chinese (Mandarin, Putonghua) and evaluate them using a defined rubric.
• Accurately identify target accents from provided audio samples.
• Apply consistent and objective judgment based on provided evaluation guidelines.

Welo Data, part of Welocalize, is a global AI data company with 500,000+ contributors delivering high-quality, ethical data to train the world’s most advanced AI systems. They are building smarter, more human AI with a diverse community in 100+ countries.

• Evaluate AI-generated French speech and text for linguistic accuracy, naturalness, and educational quality.
• Assess learner speech and writing across proficiency levels from CEFR Pre-A1 through B2+.
• Apply expert judgment to identify learner errors, unnatural phrasing, and pedagogical gaps.

Alignerr partners with leading AI labs to build expert-driven data pipelines. They improve how models reason, learn, and communicate by working with domain specialists to evaluate and refine AI systems where precision, pedagogy, and human judgment matter most.

• Review short, pre-segmented datasets.
• Evaluate model-generated replies based on Tone or Fluency .
• Read a user prompt and two model replies, then rate each using a five-point scale.

CrowdGen, by Appen, focuses on AI response evaluation. They are looking for native Javanese speakers to contribute to a multilingual AI response evaluation project where you review large language model outputs.

• Evaluate LLM-generated transcriptions on their ability to effectively understand conversations from an audio recording.
• Generate high-quality human annotations to correct model-generated transcripts.
• Apply consistent annotations by following clear taxonomies and detailed evaluation guidelines.

Neon partners with leading AI teams to improve the quality, usefulness, and reliability of audio-based AI systems. Projects focus on evaluating and improving transcriptions of audio recordings generated by large language models (LLMs).

• Evaluate AI-generated responses for accuracy, grammar, and cultural relevance.
• Identify issues and provide refined, high-quality rewritten responses.
• Create natural prompts and responses in Spanish to improve conversational datasets.

Welo Data, part of Welocalize, is a global AI data company with 500,000+ contributors delivering high-quality, ethical data to train the world’s most advanced AI systems. They're building smarter, more human AI with a diverse community in 100+ countries.

• Native or near-native fluency in Central Khmer.
• Based in: Cambodia, Thailand.
• Comfortable with digital tools.

Welo Data, part of Welocalize, is a global AI data company with 500,000+ contributors delivering high-quality, ethical data to train the world’s most advanced AI systems. They’re building smarter, more human AI with a diverse community in 100+ countries.

• Record a scripted conversation with another speaker using natural code-switching between your native language (Turkish) and English.
• Complete a short Pre-Vet test on our Aurora Studio Platform (the Pre-Vet consists of recording 4 sentences).
• If successful, you’ll be matched with another speaker and scheduled for a recording session within the next 2–3 weeks.

Lionbridge is an AI data training company. They provide opportunities to contribute to cutting-edge AI technology on a global project.

• Design scenario-based and edge-case prompts to test AI behavior.
• Develop evaluation rubrics to assess AI responses across multiple criteria.
• Perform side-by-side evaluations of AI outputs and score them using defined criteria.

Welo Data, part of Welocalize, is a global AI data company with 500,000+ contributors delivering high-quality, ethical data to train the world’s most advanced AI systems. They are building smarter, more human AI with a diverse community in 100+ countries.

• Recording spoken versions of technical scripts in fluent, natural-sounding German (Germany)
• Delivering clean, well-paced audio with high quality voice performance
• Maintaining consistency in tone, clarity, and pronunciation

Upwork has partnered with an enterprise client who is a leader in the data-centric AI space, providing a platform used to develop intelligent applications. Their technology enables teams to apply the right balance of human supervision and automation to AI development.

• Data collection, evaluation, and annotation.
• Pairwise comparisons.
• Object tagging and labeling across different content types (audio, video, images, or collected data)

RWS TrainAI focuses on improving AI-generated content. They embrace DEI and are an equal opportunity employer committed to providing a work environment free of discrimination and harassment.

• Record a scripted conversation with another speaker using natural code-switching.
• Complete a short Pre-Vet test on our Aurora Studio Platform.
• Matched with another speaker and scheduled for a recording session within the next 2–3 weeks.

Lionbridge is not described in the job posting. We improve AI speech recognition technology.

• Record short, conversational audio clips in Dutch based on a provided script.
• Simulate natural background environments such as a car, café, or office.
• Follow specific guidelines to ensure clarity, consistency, and a natural delivery.

Welocalize enables brands and companies to reach, engage, and grow international audiences. Welocalize delivers multilingual content transformation services in translation, localization, and adaptation for over 250 languages with a growing network of over 400,000 in-country linguistic resources.