tasal9/ZamAI-Whisper-v3-Pashto

ZamAI Whisper v3 Pashto

Fine-tuned Whisper model for Pashto automatic speech recognition.

Model Details

• Base Model: OpenAI Whisper
• Language: Pashto (ps)
• Task: Automatic Speech Recognition
• Fine-tuned on: Pashto speech dataset

Usage

from transformers import pipeline

transcriber = pipeline("automatic-speech-recognition", model="tasal9/ZamAI-Whisper-v3-Pashto")
result = transcriber("path_to_audio.wav")
print(result["text"])

Performance

• Word Error Rate (WER): ~15.2%
• Optimized for Afghan Pashto dialects

Training Data

Fine-tuned on diverse Pashto speech samples including:

• News broadcasts
• Conversational speech
• Educational content
• Cultural discussions

Limitations

• Primarily trained on Afghan Pashto
• Performance may vary with strong accents or background noise
• Best results with clear audio (16kHz sampling rate)