Publiée le 17 nov. 2026 · Nous avons confirmé le 17 nov. 2026 qu'elle est toujours active
Cette entreprise est la vôtre ?₹ 12.500 – ₹ 37.500 par projet
I’m building a speech dataset focused on real-world customer service calls and need help both sourcing the audio and annotating it. Each call typically features two speakers—an agent and a customer. For every recording you deliver, I require clear, time-aligned labels indicating who is speaking and when they begin and end. Full verbatim transcripts or sentiment tags aren’t necessary; the task is strictly speaker identification with accurate time stamps. You may already have compliant recordings or the ability to capture fresh ones in line with privacy regulations. Either way, I’ll need: • Clean stereo or high-quality mono audio (WAV or FLAC preferred) • A corresponding annotation file (JSON or CSV) listing start/end times and the speaker label for each segment (e.g., Agent_01, Customer_01) • Confirmation that all data is lawfully collected and free for research use I plan to review an initial sample before green-lighting a larger batch, so please outline how many minutes you can deliver in the first milestone and your typical turnaround time.
Créez un compte gratuit pour voir l'offre complète et postuler.