EngineX-Iluvatar/enginex_bi_series-sherpa-onnx
Archived
16
0
Fork 0
Code Issues Pull Requests Actions Projects Releases Wiki Activity

support whisper turbo (#1390)

Browse Source
This commit is contained in:
Fangjun Kuang
2024-10-02 18:13:34 +08:00
committed by GitHub
parent b965f14cf0
commit 66feecb2b5
2 changed files with 7 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
.github/workflows/export-whisper-to-onnx.yaml vendored
View File

@@ -16,7 +16,7 @@ jobs:
fail-fast: false fail-fast: false
matrix: matrix:
os: [macos-latest] os: [macos-latest]
model: ["distil-medium.en", "distil-small.en", "tiny.en", "base.en", "small.en", "medium.en", "tiny", "base", "small", "medium", "medium-aishell", "large", "large-v1", "large-v2", "large-v3", "distil-large-v2"] model: ["turbo", "distil-medium.en", "distil-small.en", "tiny.en", "base.en", "small.en", "medium.en", "tiny", "base", "small", "medium", "medium-aishell", "large", "large-v1", "large-v2", "large-v3", "distil-large-v2"]
# model: ["large", "large-v1", "large-v2", "large-v3", "distil-large-v2"] # model: ["large", "large-v1", "large-v2", "large-v3", "distil-large-v2"]
python-version: ["3.8"] python-version: ["3.8"]
@@ -32,7 +32,8 @@ jobs:
shell: bash shell: bash
run: | run: |
python3 -m pip install torch==1.13.0 torchaudio==0.13.0 -f https://download.pytorch.org/whl/cpu/torch_stable.html python3 -m pip install torch==1.13.0 torchaudio==0.13.0 -f https://download.pytorch.org/whl/cpu/torch_stable.html
python3 -m pip install openai-whisper==20231117 onnxruntime onnx soundfile librosa python3 -m pip install -U openai-whisper
python3 -m pip install onnxruntime onnx soundfile librosa
- name: export ${{ matrix.model }} - name: export ${{ matrix.model }}
shell: bash shell: bash

7
scripts/whisper/export-onnx.py
View File

@@ -46,7 +46,8 @@ def get_args():
choices=[ choices=[
"tiny", "tiny.en", "base", "base.en", "tiny", "tiny.en", "base", "base.en",
"small", "small.en", "medium", "medium.en", "small", "small.en", "medium", "medium.en",
"large", "large-v1", "large-v2", "large-v3", "large-v1", "large-v2",
"large", "large-v3", "turbo", # these three have feature dim 128
"distil-medium.en", "distil-small.en", "distil-large-v2", "distil-medium.en", "distil-small.en", "distil-large-v2",
# "distil-large-v3", # distil-large-v3 is not supported! # "distil-large-v3", # distil-large-v3 is not supported!
# for fine-tuned models from icefall # for fine-tuned models from icefall
@@ -76,7 +77,7 @@ def add_meta_data(filename: str, meta_data: Dict[str, Any]):
meta.key = key meta.key = key
meta.value = str(value) meta.value = str(value)
if "large" in filename: if "large" in filename or "turbo" in filename:
external_filename = filename.split(".onnx")[0] external_filename = filename.split(".onnx")[0]
onnx.save( onnx.save(
model, model,
@@ -404,7 +405,7 @@ def main():
audio = whisper.pad_or_trim(audio) audio = whisper.pad_or_trim(audio)
assert audio.shape == (16000 * 30,), audio.shape assert audio.shape == (16000 * 30,), audio.shape
if args.model in ("large", "large-v3"): if args.model in ("large", "large-v3", "turbo"):
n_mels = 128 n_mels = 128
else: else:
n_mels = 80 n_mels = 80