# Load model directly
from transformers import AutoProcessor, AutoModelForSpeechSeq2Seq
processor = AutoProcessor.from_pretrained("cdli/whisper-small_finetuned_kenyan_swahili_nonstandard_speech_v0.9")
model = AutoModelForSpeechSeq2Seq.from_pretrained("cdli/whisper-small_finetuned_kenyan_swahili_nonstandard_speech_v0.9", device_map="auto")Quick Links
Trained on CDLI's non-standard speech dataset (Kenyan Swahili): cdli/kenyan_swahili_nonstandard_speech_v0.9
Performance on test set before and after:
- before: WER on test: 0.47
- after: WER on test: 0.31
More details: https://www.cdl-inclusion.com/ASRKenya
- Downloads last month
- 4
Model tree for cdli/whisper-small_finetuned_kenyan_swahili_nonstandard_speech_v0.9
Base model
openai/whisper-small
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("automatic-speech-recognition", model="cdli/whisper-small_finetuned_kenyan_swahili_nonstandard_speech_v0.9")