Downloads · 30 days
31
48% of all-time downloads
BradUrbiz/tagatranscribe-v1
tagatranscribe-v1 is a automatic speech recognition model from BradUrbiz. Use it when you need speech turned into text. It is set up for transformers. The card lists the license as mit.
Fine-tuned OpenAI Whisper-small on 700+ Tagalog and Taglish recordings from Filipino American speakers to improve speech recognition accuracy for a low-resource language.
Downloads · 30 days
31
48% of all-time downloads
All-time downloads
64
Public
Parameters
242M
967 MB on disk
Likes
1
Public
Click a slice to open those files.
.safetensors967 MB · 99%
From the Hugging Face model README
Fine-tuned OpenAI Whisper-small on 700+ Tagalog and Taglish recordings from Filipino American speakers to improve speech recognition accuracy for a low-resource language.
TagaTranscribe-v1 is an automatic speech recognition (ASR) model created by Brad Urbiztondo as part of a STEM Fair research project investigating whether fine-tuning Whisper-small on Tagalog and Taglish audio improves transcription accuracy.
The training data was collected from willing participants in a school Filipino Cultural Club (Kasama). The dataset was created to help address the lack of speech recognition resources for Tagalog, particularly American-accented Tagalog, mixed English-Tagalog speech (Taglish), imperfect speech, and natural conversational patterns.
This model is intended for speech transcription of Tagalog and Taglish audio, especially speech from Filipino American speakers. Feel free to use however you want. Dataset is accessible on my Github repo.
The dataset was collected primarily from Filipino American speakers and may not fully represent all regional accents, dialects, speaking styles, or demographics of Tagalog speakers.
Because the dataset contains approximately 700 recordings, transcription quality may vary depending on speaker accent, audio conditions, recording quality, and vocabulary.
Use the code below to get started with the model.
from transformers import pipeline
pipe = pipeline( "automatic-speech-recognition", model="BradUrbiz/tagatranscribe-v1" )
result = pipe("audio.wav") print(result["text"])