pywhispercpp
Python bindings for whisper.cpp
Decision gist · record as of 2026-08-14
Yes, if you need local, offline speech-to-text transcription. The package is actively maintained, has no known vulnerabilities, and offers broad hardware acceleration options. Install friction is moderate due to C++ compilation, but pre-built wheels ease setup for common platforms. The MIT license imposes no restrictions. Verify that pre-built wheels include the accelerator support you need, or be prepared to build from source with environment flags.AI-flagged interpretation of the facts on this page — verify before relying
Before you install
- ffmpeg must be installed on the system to transcribe non-WAV formats (MP3, FLAC, etc.); WAV files work without it.
- For GPU acceleration, CUDA, CoreML, or other backends must be installed before building from source.
- Medium install friction due to compiled C++ bindings; pre-built wheels cover multiple Python versions on Linux, macOS, and Windows, but source builds require a C++ toolchain.
License · maintenance · safety
MIT (permissive) — MIT license is permissive; you may use, modify, and distribute pywhispercpp freely in commercial and private projects with minimal restrictions.
last release 2026-05-30 (76 days) · last repo commit 2026-07-30 · 345 stars
0 known vulnerabilities (OSV.dev, 2026-08-14) · 232,045 downloads/mo, #9,076 on PyPI
Alternatives
Verify before relying
pip install pywhispercpp
from pywhispercpp.model import Model
model = Model('base.en')
segments = model.transcribe('file.wav')
for segment in segments:
print(segment.text)- Whether pre-built wheels include CUDA, CoreML, Vulkan, or other accelerator support, or if source builds are required for GPU/specialized hardware
- Actual transcription speed and accuracy compared to other Whisper implementations
- Memory footprint and latency characteristics for different model sizes
What it is and what it does
pywhispercpp wraps the C++ implementation of OpenAI's Whisper speech recognition model, exposing it through a Python API. It downloads and caches models automatically, then transcribes audio files to text, returning segments with timing information. The package supports multiple backends (CUDA, CoreML, Vulkan, OpenBLAS, OpenVINO) via environment variables at install time, allowing optimization for different hardware. It depends on numpy, requests, tqdm, and platformdirs for core functionality.
You use it by instantiating a Model with a model name (e.g., 'base.en'), then calling transcribe() on an audio file path. The transcription runs locally without external API calls. Optional ffmpeg installation extends support beyond WAV to MP3, FLAC, and other formats. A CLI tool (pwcpp) and callback-based API enable both batch and streaming workflows.
Use it for
- Transcribe audio files offline without sending data to cloud APIs, for privacy-sensitive applications
- Build a local speech-to-text service with GPU acceleration on supported hardware platforms
- Batch-process media files to generate subtitles (SRT, VTT) or transcripts in multiple output formats
- Integrate real-time transcription into Python applications via segment callbacks during inference
- Translate non-English audio to English text using the built-in translation capability
Worth the install?
AI-flagged interpretation of the facts on this page. Verify before relying on it.
Yes, if you need local, offline speech-to-text transcription.
The package is actively maintained, has no known vulnerabilities, and offers broad hardware acceleration options. Install friction is moderate due to C++ compilation, but pre-built wheels ease setup for common platforms. The MIT license imposes no restrictions. Verify that pre-built wheels include the accelerator support you need, or be prepared to build from source with environment flags.
Install
pywhispercpp on PyPI
Before you install
Medium install friction due to compiled C++ bindings; pre-built wheels cover multiple Python versions on Linux, macOS, and Windows, but source builds require a C++ toolchain. Optional ffmpeg dependency for non-WAV formats. Active maintenance with recent release.
ffmpeg must be installed on the system to transcribe non-WAV formats (MP3, FLAC, etc.); WAV files work without it. For GPU acceleration, CUDA, CoreML, or other backends must be installed before building from source.
License in practice
MIT license is permissive; you may use, modify, and distribute pywhispercpp freely in commercial and private projects with minimal restrictions.
Quickstart
pip install pywhispercpp
from pywhispercpp.model import Model
model = Model('base.en')
segments = model.transcribe('file.wav')
for segment in segments:
print(segment.text)Verify before relying
- Whether pre-built wheels include CUDA, CoreML, Vulkan, or other accelerator support, or if source builds are required for GPU/specialized hardware
- Actual transcription speed and accuracy compared to other Whisper implementations
- Memory footprint and latency characteristics for different model sizes
Package facts
| License | MIT permissive |
| Python support | Supports the current Python release >=3.8 |
| Install friction | Medium. Platform-specific wheel |
| Runtime dependencies | 4 packagesnumpyrequeststqdmplatformdirs |
| Maintenance | Actively maintained 76 days since the last release |
| Last repo commit | |
| First released | |
| Downloads | 232,045 / month, #9,076 on PyPI 30-day window, as of 2026-08-14 |
| Known vulnerabilities | None known OSV.dev, checked 2026-08-14 |
Evidence: pywhispercpp-1.5.0-cp310-cp310-macosx_11_0_arm64.whl; pywhispercpp-1.5.0-cp310-cp310-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl; pywhispercpp-1.5.0-cp310-cp310-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl; pywhispercpp-1.5.0-cp310-cp310-musllinux_1_2_aarch64.whl; pywhispercpp-1.5.0-cp310-cp310-musllinux_1_2_x86_64.whl; pywhispercpp-1.5.0-cp310-cp310-win32.whl; pywhispercpp-1.5.0-cp310-cp310-win_amd64.whl; pywhispercpp-1.5.0-cp311-cp311-macosx_11_0_arm64.whl; pywhispercpp-1.5.0-cp311-cp311-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl; pywhispercpp-1.5.0-cp311-cp311-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl; pywhispercpp-1.5.0-cp311-cp311-musllinux_1_2_aarch64.whl; pywhispercpp-1.5.0-cp311-cp311-musllinux_1_2_x86_64.whl; pywhispercpp-1.5.0-cp311-cp311-win32.whl; pywhispercpp-1.5.0-cp311-cp311-win_amd64.whl; pywhispercpp-1.5.0-cp312-cp312-macosx_11_0_arm64.whl; pywhispercpp-1.5.0-cp312-cp312-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl; pywhispercpp-1.5.0-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl; pywhispercpp-1.5.0-cp312-cp312-musllinux_1_2_aarch64.whl; pywhispercpp-1.5.0-cp312-cp312-musllinux_1_2_x86_64.whl; pywhispercpp-1.5.0-cp312-cp312-win32.whl
Tags
Let your AI agent find packages like this
Example. Real query, live index.
You found this page by searching. An agent finds it by wishing: SkillFed indexes 14,416 PyPI packages by what they can do, searchable in plain language.
wish › “whisper.cpp python bindings”
- pywhispercpppywhispercpp provides Python bindings for whisper.cpp, enabling…
- PyQt5-sipPyQt5-sip provides the runtime support module for PyQt5 bindings,…
- PyGObjectPyGObject provides Python bindings for GObject-based libraries such…
Give your agent the search over MCP, or paste the wish link into any chat.
More Artificial Intelligence packages
LiteLLM provides a unified Python interface to call 100+ LLM providers (OpenAI, Anthropic, Gemini, Bedrock, Azure, and others) using OpenAI-compatible API format, available as both a Python SDK and a self-hosted AI Gateway proxy server.
Install it if you need to work with multiple LLM providers or want to centralize LLM routing in your organization.
Client library and CLI tool for downloading, uploading, and managing models, datasets, and repositories on the Hugging Face Hub platform.
Install it if you work with Hugging Face Hub models or datasets.
LangChain provides a framework for building agents and LLM-powered applications by composing language models, tools, and memory through a unified API that abstracts over multiple model providers.
hf-xet provides chunk-based deduplication and efficient file transfer for the Hugging Face Hub, enabling faster uploads and downloads of large files with local disk caching.
Tokenizers converts raw text into token sequences for NLP models, with support for training custom vocabularies and using pre-built tokenizers (BPE, WordPiece) optimized for speed via Rust.
Transformers provides a unified framework for loading, fine-tuning, and running state-of-the-art pretrained models across text, vision, audio, video, and multimodal tasks using PyTorch, JAX, or TensorFlow.
Install it if you need to run or train any transformer-based model for NLP, vision, audio, or multimodal tasks.
See also faster-whisper · openai-whisper · SpeechRecognition · realtimestt · whisperx · whisper-timestamped · voxcpm · gTTS · pymicro-vad · llama-cpp-python