Native Japanese Speaker for AI-Generated Speech Evaluation（日本語ネイティブ音声評価）

We are building a Japanese-language evaluation team to review synthetic speech generated by AI voice systems. This is a paid, remote project focused on the linguistic naturalness and audio quality of Japanese speech. Evaluators who consistently provide accurate and reliable judgments may be invited to participate in additional Japanese evaluation batches. Your responsibilities Listen to short AI-generated Japanese speech samples through our secure evaluation workspace and assess their naturalness, intelligibility, pronunciation, prosody, rhythm, and overall audio quality. Pay particular attention to Japanese-specific features such as pitch accent, long vowels, small っ, ん, and mora timing. Check whether kanji and counter words are read correctly, whether sound changes such as rendaku (連濁) sound natural, whether words and phrases are segmented appropriately, and whether polite or plain speech fits the context. Compare multiple speech samples according to written evaluation guidelines and identify the stronger version. Mark specific words or timestamps when an issue needs to be localized. Promptly report technical problems or unclear task instructions. Occasionally complete a small number of short Japanese voice-recording tasks by following written prompts. Recording will represent only a small portion of the overall workload. Participate in a short paid debrief interview after completing an evaluation batch when requested. Who we are looking for Japanese is your first language. You acquired Japanese primarily in Japan and spent most of your childhood there, particularly before age 12. You are comfortable evaluating standard Japanese/common Japanese (共通語) as used in professional media and everyday communication. Applicants from different regions of Japan are welcome, but you should be able to distinguish regional speech patterns from intended standard Japanese. You can explain your judgments clearly in written English. You can commit approximately 1–5 hours per week and respond to or complete assigned tasks within 24 hours of receiving them. Equipment and working environment Reliable headphones or earphones. A quiet environment suitable for careful listening and occasional recording. A reliable computer and internet connection. The ability to follow structured evaluation guidelines consistently. Relevant experience Experience in speech evaluation, language annotation, transcription, localization, voice recording, audio quality assurance, or language teaching is helpful but not required. A linguistics degree is not required. Role separation and conflicts Please tell us if you have previously written, reviewed, translated, or recorded any material connected with this project. You will not be assigned content that you helped create. Selection process Submit complete answers to the screening questions. Attend a 15–30 minute Upwork interview covering your Japanese-language background, equipment, availability, and rate. Selected candidates will receive an Upwork contract and complete a small, paid first batch. Candidates whose first batch meets our quality requirements may continue to additional Japanese evaluation batches. Training and compensation We will provide one free introductory training session before the first task. The training session is informational and unpaid. It will not require you to submit any evaluation, recording, or other usable project output. Evaluation rate: USD 15–20 per hour. All assigned evaluation, voice-recording, and debrief work after the contract begins is paid. The recruitment interview is unpaid and will not require a usable work sample. Any reading, recording, or evaluation sample that may be used by the project will be assigned and paid only after the contract begins. Please apply with complete and accurate answers to all screening questions.