🎙 LEHJA — Urdu voice-model training (live)

Updated: 2026-06-12 17:00:01 UTC · cron har 5 min · page 60s auto-refresh

OVERALL PIPELINE: 38%
GPU: 100% busy · jobs: 17 · VRAM: 69.1 GB used / 23.9 GB free / 93.6 GB total

📋 Tasks (main + sub)

1. H100 Setup ✅ 100%
   ✅ SSH access + box verification — complete (2 min lagi)
   ✅ Python deps + CosyVoice repo install — complete (6 min lagi)
   ✅ CosyVoice2-0.5B model download (5.3 GB) — complete (5 min lagi)
   ✅ Whisper CUDA-12 libs fix (silent-fail pakra gaya tha) — complete (14 min lagi)
2. Data Acquisition ✅ 100%
   ✅ Catalogued Urdu files VM101→H100 (2,667 files) — complete (3 min lagi)
   ✅ Full pool transfer (93,166 recordings, 17 GB) — complete (9 min lagi)
   ✅ Catalog language-map (45,185 entries) — complete (1 min lagi)
   ✅ Public FLEURS Urdu+English — 3,637 clips — complete (26 min lagi)
   ⊘ CommonVoice Urdu (HF-gated, zaroorat nahi) — skipped
3. In-domain Urdu Mining (asal awaazein) 🔄 58%
   🔄 Known-Urdu transcription — 2,667 files, 6 GPU workers — 64% · 54 min se chal raha · ~30 min baqi · 1,700/2,668 files · 10,640 clips · 558.9 min saaf Urdu
   ✅ Language-scan — 53,054 unknown files, 16 GPU workers — complete (18 min lagi) · 53,054/53,054 scanned · naye Urdu-candidates: 3,219
   🔄 Round-2: naye mile files ki transcription — 31% · 65 min se chal raha · ~2.4 h baqi · 1,000/3,219
4. Training-Data Prep ⬜ 0%
   ⬜ Manifests merge + dedup + train/dev split — pending
   ⬜ wav.scp / text / utt2spk / spk2utt — pending
   ⬜ Speaker embeddings (campplus) — pending
   ⬜ Speech tokens extraction — pending
   ⬜ Parquet packaging — pending
5. CosyVoice2 URDU TRAINING ⬜ 0%
   ⬜ LLM module fine-tune (Urdu phonetics) — pending · ~10 epochs planned
   ⬜ Checkpoint averaging (best-5) — pending
   ⬜ Inference smoke-test (Urdu bolta hai?) — pending
6. Validation ⬜ 0%
   ⬜ Urdu test-set synthesis (50 jumlay) — pending
   ⬜ Whisper-readback intelligibility gate — pending
   ⬜ Zero-shot voice-clone spot test (apki awaaz) — pending
7. Delivery ⬜ 0%
   ⬜ Sab artifacts app-server par download (model+data+logs) — pending
   ⬜ COMPLETION EMAIL → info@ifcondition.com → AAP H100 BAND KAR DEN — pending

📊 Harvest

Saaf Urdu (training-ready): 558.9 min in-domain (10,640 clips) + 3,637 public clips
Scan se naye Urdu-candidate files: 3,219 / 53,054 scanned · worker failures: 0

Kaam jari hai — mukammal hone par email: info@ifcondition.com