A float16 Core ML conversion of Oriserve/Whisper-Hindi2Hinglish-Prime packaged for WhisperKit on Apple Silicon (macOS 14+ / iOS 17+).
It ships compiled .mlmodelc bundles, ready for on-device inference through WhisperKit's Swift API. For Python inference, use the upstream repo linked above.
Output script: romanized Hinglish (Latin script). Hindi speech comes back transliterated into English letters, not Devanagari.
import WhisperKit
let pipe = try await WhisperKit(
WhisperKitConfig(
model: "Oriserve_Whisper-Hindi2Hinglish-Prime_fp16",
modelRepo: "vbhar/whisperkit-hindi2hinglish-prime-coreml"
)
)
let result = try await pipe.transcribe(audioPath: "meeting.wav")
Or download the variant folder directly and point WhisperKit at a local path:
hf download vbhar/whisperkit-hindi2hinglish-prime-coreml \
--include "Oriserve_Whisper-Hindi2Hinglish-Prime_fp16/*" \
--local-dir ./models
Oriserve_Whisper-Hindi2Hinglish-Prime_fp16/
├── MelSpectrogram.mlmodelc/
├── AudioEncoder.mlmodelc/ # ~1.2 GB weights
├── TextDecoder.mlmodelc/ # ~1.7 GB weights
├── config.json
└── generation_config.json
No quantized variants and no decoder prefill data are included — float16 only, by choice (see A note on quantization below).
Lineage:
| This repo | vbhar/whisperkit-hindi2hinglish-prime-coreml — Core ML float16 conversion |
| Direct source | Oriserve/Whisper-Hindi2Hinglish-Prime — Hinglish fine-tune, Apache-2.0 |
| Upstream architecture | openai/whisper-large-v3 |
All credit for the model itself belongs to Oriserve. Licensed under the Apache License, Version 2.0; see https://www.apache.org/licenses/LICENSE-2.0.
Statement of modifications (Apache-2.0 §4(b)): the weights were converted from float32 to float16 and compiled to Core ML for the Apple Neural Engine. They are otherwise unmodified — no fine-tuning, no distillation, no pruning, no vocabulary change, no re-training. The only non-weight change made during conversion was supplying alignment_heads metadata that the source repo ships as null (see Conversion gotcha below); those heads are the standard whisper-large-v3 values and affect word-timestamp attention selection only, not transcription weights.
Converted with argmaxinc/whisperkittools under Python 3.11 (3.14 has no working coremltools/torch wheels as of this writing):
python3.11 -m venv .venv && source .venv/bin/activate
pip install git+https://github.com/argmaxinc/whisperkittools.git
whisperkit-generate-model \
--model-version Oriserve/Whisper-Hindi2Hinglish-Prime \
--output-dir out \
--repo-path-suffix fp16
float16, no quantized variants, no decoder prefill data.
alignment_heads is nullOriserve/Whisper-Hindi2Hinglish-Prime ships generation_config.json with:
"alignment_heads": null
The whisperkittools decoder trace uses alignment_heads to build the alignment_heads_weights output. With null, the traced decoder emits 3 outputs while the conversion declares 4, and the run dies with:
ValueError: Number of outputs provided, 4, do not match the number of outputs
produced by the model, 3
Fix: fall back to the standard whisper-large-v3 alignment heads — 10 [layer, head] pairs — before tracing:
generation_config.alignment_heads = [
[7, 0], [10, 17], [12, 18], [13, 12], [16, 1],
[17, 14], [19, 11], [21, 4], [24, 1], [25, 6],
]
This is the correct fallback, not a guess: the model is an unmodified-architecture whisper-large-v3 fine-tune, and these pairs are byte-identical to the ones the sibling vbhar/whisperkit-hinglish-large-v3-coreml build carries from its own source. The values are baked into the generation_config.json in this repo, so a consumer of this repo never sees the failure — only someone re-converting from the Oriserve source does.
Float16 was chosen deliberately. During this work, a quantized Core ML build of
these weights was observed to emit the literal string nan on real
code-switched Hindi/English speech, while passing cleanly on short English
clips such as WhisperKit's bundled jfk.wav.
The asymmetry is the point worth noting: quantization error is input-dependent, so a short smoke test — the normal way to sanity-check a conversion — can pass on a build that fails on the audio you actually care about. If you evaluate a quantized Whisper conversion, test it on your own code-switched audio at realistic length, not on a short clip.
No quantized variant is published here, because none was validated on that kind of input.
vbhar/whisperkit-hinglish-large-v3-coreml instead. Picking the wrong one of the two is the most likely mistake with this pair.Tested live end-to-end in a real macOS menu-bar meeting-notes app (mic + system audio, streaming transcription) on natural Hindi/English code-switched speech.
Hindi is returned transliterated into Latin script and English is returned as-is, mixed inline within the same sentence.
Oriserve_Whisper-Hindi2Hinglish-Prime_fp16/TextDecoder.mlmodelc/weights/weight.bin 1.8 GBOriserve_Whisper-Hindi2Hinglish-Prime_fp16/AudioEncoder.mlmodelc/weights/weight.bin 1.3 GBOriserve_Whisper-Hindi2Hinglish-Prime_fp16/MelSpectrogram.mlmodelc/weights/weight.bin 373.4 kBOriserve_Whisper-Hindi2Hinglish-Prime_fp16/TextDecoder.mlmodelc/coremldata.bin 0.7 kBOriserve_Whisper-Hindi2Hinglish-Prime_fp16/AudioEncoder.mlmodelc/coremldata.bin 0.4 kBOriserve_Whisper-Hindi2Hinglish-Prime_fp16/MelSpectrogram.mlmodelc/coremldata.bin 0.4 kBOriserve_Whisper-Hindi2Hinglish-Prime_fp16/AudioEncoder.mlmodelc/analytics/coremldata.bin 0.2 kBOriserve_Whisper-Hindi2Hinglish-Prime_fp16/MelSpectrogram.mlmodelc/analytics/coremldata.bin 0.2 kBOriserve_Whisper-Hindi2Hinglish-Prime_fp16/TextDecoder.mlmodelc/analytics/coremldata.bin 0.2 kBHave the files? Submit a torrent.
Submit a torrentHave the complete files and permission to share them? Replace the example folder path. The script checks the files, makes a torrent and submits its magnet and checksums. Keep your torrent client seeding. Get your community key from your settings, and read the script before running it.
curl -fsSL https://pirateface.co/package.sh | bash -s -- --repo vbhar/whisperkit-hindi2hinglish-prime-coreml ./model-folderOriserve/Whisper-Hindi2Hinglish-Prime