whisper-finetuned-float32

AfroLogicInsect/whisper-finetuned-float32

Fine-tuned Whisper model (float32 version) for speech recognition

Model Details

• Model Type: Whisper (Fine-tuned)
• Language: English
• Data Type: float32
• Use Cases: Speech-to-text transcription

Usage

from transformers import WhisperProcessor, WhisperForConditionalGeneration
import librosa

# Load model and processor
processor = WhisperProcessor.from_pretrained("AfroLogicInsect/whisper-finetuned-float32")
model = WhisperForConditionalGeneration.from_pretrained("AfroLogicInsect/whisper-finetuned-float32")

# Load audio
audio, sr = librosa.load("path/to/audio.wav", sr=16000)

# Process
input_features = processor(audio, sampling_rate=16000, return_tensors="pt").input_features

# Generate transcription
with torch.no_grad():
    predicted_ids = model.generate(input_features)
    transcription = processor.batch_decode(predicted_ids, skip_special_tokens=True)[0]

print(transcription)

Training Details

• Base Model: OpenAI Whisper
• Training Dataset: [Add your dataset details]
• Training Parameters: [Add your training parameters]
• Evaluation Metrics: [Add your evaluation results]

Limitations and Biases

• This model may have biases present in the training data
• Performance may vary on different accents or audio qualities
• Recommended for English speech recognition tasks

Citation

If you use this model, please cite:

@misc{whisper-finetuned,
  author = {Daniel AMAH},
  title = {Fine-tuned Whisper Model},
  year = {2024},
  publisher = {Hugging Face},
  url = {https://huggingface.co/AfroLogicInsect/whisper-finetuned-float32}
}