Add Intel® OpenVINO Whisper large v3 turbo model

[Anirudh-Swaminathan]

• Introduced audio_processor_config_default.json for audio feature extraction configuration.
• Created convert_whisper_to_ovir.py for converting Whisper models to OpenVINO format using Olive.
• Added info.yml to define model architecture and recipes for OpenVINO execution.
• Implemented whisper_large_v3_turbo_default_ov_npu.json for model conversion configuration targeting NPU.
• Developed whisper_large_v3_turbo_encapsulate.json for encapsulating the converted models with necessary settings.

---

[Copilot]

Pull request overview

Adds an Intel® OpenVINO NPU conversion + encapsulation recipe for openai/whisper-large-v3-turbo, including an Olive-based conversion script, model metadata, and audio feature-extraction configuration. Also updates several OpenVINO-related requirement pins and bumps Intel NPU toolchain versions.

Changes:

• Add OpenVINO NPU Olive recipe configs + metadata (info.yml) for Whisper large v3 turbo.
• Add convert_whisper_to_ovir.py helper to run conversion + encapsulation and post-process genai_config.json.
• Update OpenVINO/IntelNPU dependency versions and loosen some model-specific OpenVINO requirements.

Reviewed changes

Copilot reviewed 11 out of 11 changed files in this pull request and generated 5 comments.

Show a summary per file

File	Description
openai-whisper-large-v3-turbo/OpenVINO/whisper_large_v3_turbo_encapsulate.json	Adds encapsulation pass config for OpenVINO NPU (genai override + provider options).
openai-whisper-large-v3-turbo/OpenVINO/whisper_large_v3_turbo_default_ov_npu.json	Adds Olive Optimum conversion + IO update config targeting NPU.
openai-whisper-large-v3-turbo/OpenVINO/info.yml	Registers the OpenVINO recipe for discovery/automation.
openai-whisper-large-v3-turbo/OpenVINO/convert_whisper_to_ovir.py	New conversion/encapsulation driver script with optional reshape and NPU weight sharing.
openai-whisper-large-v3-turbo/OpenVINO/audio_processor_config_default.json	Adds default audio feature-extraction pipeline configuration.
openai-whisper-large-v3-turbo/OpenVINO/README.md	Documents usage and expected output artifacts for the OpenVINO pipeline.
microsoft-Phi-4/OpenVINO/requirements.txt	Loosens Optimum-Intel and Transformers requirements.
microsoft-Phi-4-reasoning/OpenVINO/requirements.txt	Loosens Optimum-Intel and Transformers requirements.
microsoft-Phi-4-mini-reasoning/OpenVINO/requirements.txt	Loosens Optimum-Intel and Transformers requirements.
microsoft-Phi-4-mini-instruct/OpenVINO/requirements.txt	Loosens Optimum-Intel and Transformers requirements.
.aitk/requirements/requirements-IntelNPU.txt	Bumps OpenVINO/Olive/ONNX/ORT stack versions for Intel NPU environment.

Comments suppressed due to low confidence (1)

openai-whisper-large-v3-turbo/OpenVINO/convert_whisper_to_ovir.py:7

• sys is imported but never used in this script. Please remove the unused import to avoid lint warnings and keep dependencies clear.

import sys
import shutil

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.