stcoats/whisper-large-v3-DASS2019-ct2

whisper-large-v3-DASS2019-ct2

Model Description

This repository contains a CTranslate2-converted version of a Whisper large-v3 model fine-tuned on DASS2019 speech data.

The model is intended for inference using:

• —`faster-whisper`
• —`whisperx`

This format is optimized for:

• —fast GPU inference
• —reduced memory footprint
• —production ASR pipelines

Base Model

Fine-tuned from:

openai/whisper-large-v3

Converted to CTranslate2 format for inference acceleration.

Intended Use

This model is designed for:

• —Automatic speech recognition (ASR) of historical Southern American English and African-American English
• —Research transcription pipelines
• —Large-scale batch transcription
• —WhisperX alignment / diarization workflows

Out-of-Scope Use

This model is not suitable for:

• —Real-time low-latency streaming without additional engineering
• —Use via Hugging Face Transformers API
• —Applications requiring multilingual robustness (trained primarily on English speech)

How to Use

from huggingface_hub import snapshot_download
from faster_whisper import WhisperModel

model_dir = snapshot_download("stcoats/whisper-large-v3-DASS2019-ct2")
model = WhisperModel(model_dir, device="cuda", compute_type="float16")

segments, info = model.transcribe("audio.wav", language="en", beam_size=5)

for s in segments:
    print(f"[{s.start:.2f}-{s.end:.2f}] {s.text}")

or

from huggingface_hub import snapshot_download
import whisperx

model_dir = snapshot_download("stcoats/whisper-large-v3-DASS2019-ct2")
model = whisperx.load_model(model_dir, device="cuda", compute_type="float16")

audio = whisperx.load_audio("audio.wav")
result = model.transcribe(audio, language="en", vad_filter=False)

for s in result["segments"]:
    print(f"[{s['start']:.2f}-{s['end']:.2f}] {s['text']}")

Details

Training Data

DASS2019_NLP

For details, please see Coats, Steven. (2026). A Fine-tuned ASR Model for Historical American Dialect Recordings. Proceedings of LREC 2026.

Citation

BibTeX:

@inproceedings{coats-2026-fine,
  title = {A Fine-tuned ASR Model for Historical American Dialect Recordings},
  author = {Coats, Steven},
  booktitle = {Proceedings of the Fifteenth Language Resources and Evaluation Conference (LREC 2026)},
  month = {May},
  year = {2026},
  pages = {1372--1381},
  address = {Palma, Mallorca, Spain},
  publisher = {European Language Resources Association (ELRA)},
  editor = {Piperidis, Stelios and Bel, Núria and van den Heuvel, Henk and Ide, Nancy and Krek, Simon and Toral, Antonio},
  doi = {10.63317/5bjeqct6ozd3},
}

APA:

Coats, S. (2026). A Fine-tuned ASR Model for Historical American Dialect Recordings. In In S. Piperidis, N. Bel, H. van den Heuvel, N. Ide, S. Krek and A. Toral (eds.) Proceedings of the Fifteenth Language Resources and Evaluation Conference (LREC 2026) (pp. 1372–1381). European Language Resources Association (ELRA). https://doi.org/10.63317/5bjeqct6ozd3.