Update README.md

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.start, segment.end, segment.text))

README.md

@@ -1,3 +1,8 @@
 ---
 license: mit
----
+language:
+- uz
+base_model:
+- openai/whisper-medium
+pipeline_tag: automatic-speech-recognition
+---