omnifeed Search

whisper.cpp

C/C++ implementation of OpenAI's Whisper speech recognition model, with a command-line tool

Visit github.com

In our rankings

Scores on the offline transcription apps criteria

Evaluated Nov 5, 2026
  • What leaves your computer ×3 4

    4/5 means: Transcribes locally, with no online features documented

    A local library with no online features documented; the models are a separate download and a full no-internet setup is not documented.

    3 quotes
  • Speaker labels and review ×3 2

    2/5 means: Experimental speaker-change markers only

    Only experimental speaker-change markers through tinydiarize, and no transcript viewer, since it is a command-line tool.

    1 quote
  • Export formats ×2 4

    4/5 means: Four or more formats, including SRT and WebVTT

    Six output formats from the command-line tool (text, VTT, SRT, LRC, CSV, JSON), but no word-processor format.

    6 quotes
  • GPU support and platforms ×2 5

    5/5 means: As 4, on Windows, macOS and Linux, with a documented processor-only mode

    Apple Silicon through Metal and Core ML, NVIDIA, AMD ROCm and cross-vendor Vulkan, plus CPU-only inference, on macOS, Linux and Windows.

    8 quotes

Evidence (20 quotes)

  1. 1
    “Plain C/C++ implementation without dependencies”
    github.com/ggml-org/whisper.cpp · checked
  2. 2
    “As an example, here is a video of running the model on an iPhone 13 device - fully offline, on-device”
    github.com/ggml-org/whisper.cpp · checked
  3. 3
    “Then, download one of the Whisper models converted in ggml format.”
    github.com/ggml-org/whisper.cpp · checked
  4. 4
    “Speaker segmentation via tinydiarize (experimental)”
    github.com/ggml-org/whisper.cpp · checked
  5. 5
    “output result in a text file”
    github.com/ggml-org/whisper.cpp/blob/master/examples/cli/README.md · checked
  6. 6
    “output result in a vtt file”
    github.com/ggml-org/whisper.cpp/blob/master/examples/cli/README.md · checked
  7. 7
    “output result in a srt file”
    github.com/ggml-org/whisper.cpp/blob/master/examples/cli/README.md · checked
  8. 8
    “output result in a lrc file”
    github.com/ggml-org/whisper.cpp/blob/master/examples/cli/README.md · checked
  9. 9
    “output result in a CSV file”
    github.com/ggml-org/whisper.cpp/blob/master/examples/cli/README.md · checked
  10. 10
    “output result in a JSON file”
    github.com/ggml-org/whisper.cpp/blob/master/examples/cli/README.md · checked
  11. 11
    “Apple Silicon first-class citizen - optimized via ARM NEON, Accelerate framework, Metal and Core ML”
    github.com/ggml-org/whisper.cpp · checked
  12. 12
    “Efficient GPU support for NVIDIA”
    github.com/ggml-org/whisper.cpp · checked
  13. 13
    “AMD ROCm GPU support”
    github.com/ggml-org/whisper.cpp · checked
  14. 14
    “Cross-vendor solution which allows you to accelerate workload on your GPU.”
    github.com/ggml-org/whisper.cpp · checked
  15. 15
    “Support for CPU-only inference”
    github.com/ggml-org/whisper.cpp · checked
  16. 16
    “Mac OS (Intel and Arm)”
    github.com/ggml-org/whisper.cpp · checked
  17. 17
    “Linux / FreeBSD”
    github.com/ggml-org/whisper.cpp · checked
  18. 18
    “Windows (MSVC and MinGW)”
    github.com/ggml-org/whisper.cpp · checked
  19. 19
    “Note that the whisper-cli example currently runs only with 16-bit WAV files, so make sure to convert your input before running the tool.”
    github.com/ggml-org/whisper.cpp · checked
  20. 20
    “Now build the whisper-cli example and transcribe an audio file like this:”
    github.com/ggml-org/whisper.cpp · checked

Alternatives