---
license: apache-2.0
language: en
tags:
- automatic-speech-recognition
- whisper
- indian-english
base_model: openai/whisper-tiny
library_name: transformers
pipeline_tag: automatic-speech-recognition
---

# Whisper Tiny: fine-tuned for Indian English

Fine-tuned [openai/whisper-tiny](https://huggingface.co/openai/whisper-tiny) (39M params) on
[TIE_shorts](https://huggingface.co/datasets/raianand/TIE_shorts), a corpus of NPTEL-style
Indian-English academic lecture audio. Part of [Indian-ASR-Bench](https://github.com/theshivam7/indian-asr-bench),
a reproducible WER benchmark for ASR on Indian English speech.

## Usage

```python
from transformers import pipeline

pipe = pipeline("automatic-speech-recognition", model="theshivam7/whisper-tiny-indian-english")
result = pipe("audio.wav")
print(result["text"])
```

For audio longer than 30 seconds, enable chunking:

```python
result = pipe("long_audio.wav", chunk_length_s=30, stride_length_s=5)
```

## Model description

A full fine-tune (all 39M parameters) of Whisper Tiny via `transformers` `Seq2SeqTrainer`,
using a step-based recipe (`max_steps=2000`, effective batch size 32, fp16, best-checkpoint
selection on validation WER). Best checkpoint: step 600/2000.

## Intended uses & limitations

Evaluated on Indian-accented English academic lecture speech (TIE_shorts). Not evaluated on
other English varieties or domains.

> ⚠️ **Speaker overlap (disclosed).** The training data's official split shares speakers with
> the test split (100% of test speakers, and 100% of test clips, are from speakers also seen in
> training). There is no clip-level leakage, but part of the measured gain over the pretrained
> baseline likely reflects speaker adaptation rather than purely accent/content adaptation. See
> the [benchmark repo](https://github.com/theshivam7/indian-asr-bench) for full methodology.

> ⚠️ **Gain is not statistically confirmed.** The −2.96pp WER improvement over the pretrained
> baseline (below) does not survive Holm–Bonferroni correction across the 3-model fine-tuning
> family at 985 test clips (95% CI [−6.35, +0.13] pp, p<sub>Holm</sub>=0.195); read it as
> suggestive, not conclusive.

## Training and evaluation data

- **Train**: [TIE_shorts](https://huggingface.co/datasets/raianand/TIE_shorts) official train
  split, 7,200 filtered clips (46.9h).
- **Eval**: TIE_shorts official test split, 985 scored clips (one excluded for an empty
  reference).

## Training procedure

| | |
|---|---|
| Base model | openai/whisper-tiny (39M) |
| Steps | 2000 (best checkpoint at step 600) |
| Effective batch size | 32 |
| Precision | fp16 |
| Optimizer | AdamW (`transformers` `Seq2SeqTrainer` default) |
| Selection | best checkpoint by validation WER |

## Evaluation results

Corpus WER on TIE_shorts test split, `transcript_clean` normalization, both models decoded
through the identical HF `transformers` pipeline (engine-controlled comparison):

| Model | Corpus WER |
|---|:---:|
| openai/whisper-tiny (pretrained baseline) | 22.10% |
| **theshivam7/whisper-tiny-indian-english (this model)** | **19.14%** |

Δ = −2.96 pp (95% CI [−6.35, +0.13], p<sub>Holm</sub>=0.195, not significant after correction,
see limitations above).

## Framework versions

- Transformers, PyTorch, Datasets (see training environment)

## Citation

Part of [Indian-ASR-Bench](https://github.com/theshivam7/indian-asr-bench). See the repo for
full benchmark methodology, error analysis, and reproduction instructions.