Urdu Voice Dataset Recording
Budget: $8 – $15 USD
I need a clean, studio-quality collection of Urdu voice recordings that I can later use for a variety of natural-language projects. Every file must follow these strict technical parameters:
• WAV, uncompressed PCM
• 44.1 kHz, 16-bit, mono
• SNR ≥ 35 dB, absolutely no clipping
• 0.5–1 second of silence at the head and tail
You will receive scripted text spanning health, news, agriculture, environmental topics and any additional domains I send along the way—my goal is to build as broad a corpus as possible. For each script, return an individual file that meets all the above specifications and is clearly named so I can trace it back to the source text.
The ideal flow is simple: I share batches of sentences, you record them in a controlled acoustic setting, run a quick QA pass to verify SNR and clipping, then hand back the final WAVs. If you have automated tools for measuring noise floor and head-room, mention them; otherwise manual checks are fine as long as the thresholds are respected.
I will review a small sample first to confirm quality before we proceed to the full set.
• WAV, uncompressed PCM
• 44.1 kHz, 16-bit, mono
• SNR ≥ 35 dB, absolutely no clipping
• 0.5–1 second of silence at the head and tail
You will receive scripted text spanning health, news, agriculture, environmental topics and any additional domains I send along the way—my goal is to build as broad a corpus as possible. For each script, return an individual file that meets all the above specifications and is clearly named so I can trace it back to the source text.
The ideal flow is simple: I share batches of sentences, you record them in a controlled acoustic setting, run a quick QA pass to verify SNR and clipping, then hand back the final WAVs. If you have automated tools for measuring noise floor and head-room, mention them; otherwise manual checks are fine as long as the thresholds are respected.
I will review a small sample first to confirm quality before we proceed to the full set.