Go See All Bilingual Jobs
Upwork Logo

Upwork

Apply
-
United States
Remote
Posted May 1, 2026

English

Cantonese

Upwork Is Hiring A High Quality Youtube Podcast Playlist Collection

We are developing a comprehensive dataset for AI-powered voice (text-to-speech) training and are seeking meticulous freelancers to source high-quality YouTube podcast playlists that adhere to strict criteria.

📝 Job Description

Join our team to identify and submit YouTube playlists featuring engaging, natural conversations suitable for AI training. This role requires careful attention to detail and strict adherence to guidelines to ensure the quality and consistency of the collected data.

🔎 What You’ll Do

  • Find and submit YouTube playlists containing:
    • Long-form conversational podcasts
    • Interviews or small group discussions
  • Ensure all submissions meet the validation criteria.

✔️ Validation Criteria

Content Type

  • Podcasts, interviews, or conversational discussions only
  • 2 speakers (interviews) or small group (max ~4 active speakers)
  • Natural, unscripted conversation

Episode Length

  • Average episode length ≥ 40 minutes
  • Exclude playlists with many episodes under 20 minutes

Audio Quality

  • Clear, intelligible speech
  • Minimal background noise
  • No distortion, clipping, or heavy echo
  • Intro/outro music acceptable only if brief
  • No continuous background music

Consistency

  • Same type of content throughout the playlist
  • ≥90% in the target language
  • No mixed or random content

Production Quality (Preferred)

  • Visible speakers or clear podcast setup
  • Stable recording environment
  • Uniform audio quality across episodes

🚫 Disallowed Content

  • Audiobooks or narration (single speaker reading)
  • Language learning or “how to speak” content
  • News broadcasts or TV clips
  • Movies or TV content
  • Religious sermons
  • Stand-up comedy

🎙️ Audio Issues to Avoid

  • Background music throughout
  • Live audience noise (clapping, cheering)
  • Poor audio quality (muffled, noisy, distorted)
  • Frequent overlapping speech

🔧 Structure Issues to Exclude

  • Single-speaker monologues
  • Highly edited or artificial audio (sound effects, jump cuts)
  • Compilation videos
  • Unrelated videos in playlists

🎯 Quality Requirements

  • Clear turn-taking between speakers
  • Minimal crosstalk
  • Calm, structured conversations
  • Consistent audio quality across episodes

📝 Submission Requirements

Details will be provided via spreadsheet, including:

  • Playlist link
  • Playlist ID
  • Channel name
  • Presence of multiple speakers (yes/no)
  • Number of episodes
  • No duplicate entries

✅ Final Check Before Submission

  • Audio is consistently clean and clear
  • Conversations are natural and unscripted
  • Content aligns with the specified requirements

If there is any uncertainty about a playlist meeting the guidelines, please exclude it from submissions. 🔍

Apply