# whisper.cpp privacy rating

C/C++ port of OpenAI's Whisper speech recognition model that runs transcription on the CPU or GPU of the local device. Used as a library and command-line tool, and as the engine inside many dictation apps.

## Summary

whisper.cpp scores 87 out of 100 (grade B) on the dictation and transcription criteria. It meets 5 of 6 criteria: open source, no trackers or telemetry, no ads or data sales, runs on the device and no training on recordings. It does not meet independent audit.

- Website: https://github.com/ggml-org/whisper.cpp
- Source code: https://github.com/ggml-org/whisper.cpp
- License: MIT
- Platforms: Windows, macOS, Linux
- Category: [Dictation and transcription](https://privacyratings.com/speech-to-text/)
- Grade: B (87/100)

## Criteria

| Criterion | Answer | Note | Evidence |
| --- | --- | --- | --- |
| Open source | Yes | MIT license. | https://github.com/ggml-org/whisper.cpp/blob/master/LICENSE |
| No trackers or telemetry | Yes | No telemetry or analytics in the source code. | https://github.com/ggml-org/whisper.cpp |
| No ads or data sales | Yes | Free open-source software with no ads. | https://github.com/ggml-org/whisper.cpp |
| Independent audit | No | No independent audit is published. |  |
| Runs on the device | Yes | Inference runs entirely on the local device. | https://github.com/ggml-org/whisper.cpp |
| No training on recordings | Yes | Processing is entirely local. | https://github.com/ggml-org/whisper.cpp |

Source: https://privacyratings.com/speech-to-text/whisper-cpp/
Edit: https://github.com/privacyratings/privacyratings.com/edit/main/ratings/speech-to-text/whisper-cpp.md