Model Description
Bu model OpenAI Whisper Medium arxitekturasiga asoslangan bo'lib, maxsus audiodata to'plamida fine-tune qilingan. Model Faster-Whisper formatiga o'tkazilgan, bu esa xotira (VRAM) sarfini kamaytirib, transkripsiya tezligini sezilarli darajada oshiradi.
Asosiy model: openai/whisper-medium
O'qitish davomiyligi: 50 soat (GPU vaqti)
Dataset hajmi: 105 soat
Epochlar: 3
Format: CTranslate2 (Faster-Whisper)
Training Results
Modelning samaradorligi quyidagi ko'rsatkichlar bilan tasdiqlangan:
WER (Word Error Rate): 16.6%
Usage (Foydalanish)
Ushbu modelni ishlatish uchun faster-whisper kutubxonasi o'rnatilgan bo'lishi kerak:
Model Description
Bu model OpenAI Whisper Medium arxitekturasiga asoslangan bo'lib, maxsus audiodata to'plamida fine-tune qilingan. Model Faster-Whisper formatiga o'tkazilgan, bu esa xotira (VRAM) sarfini kamaytirib, transkripsiya tezligini sezilarli darajada oshiradi.
Asosiy model: openai/whisper-medium
O'qitish davomiyligi: 50 soat (GPU vaqti)
Dataset hajmi: 105 soat
Epochlar: 3
Format: CTranslate2 (Faster-Whisper)
Training Results
Modelning samaradorligi quyidagi ko'rsatkichlar bilan tasdiqlangan:
WER (Word Error Rate): 16.6%
Usage (Foydalanish)
Ushbu modelni ishlatish uchun faster-whisper kutubxonasi o'rnatilgan bo'lishi kerak:
pip install faster-whisper
from faster_whisper import WhisperModel
model_path = "Chingiz5408/Faster-whisper"
# Modelni GPU yoki CPU uchun yuklash
model = WhisperModel(model_path, device="cuda", compute_type="float16")
segments, info = model.transcribe("audio.mp3", beam_size=5)
for segment in segments:
print("[%.2fs -> %.2fs] %s" % (segment…
