Whisper (GGML)
This is a mirror for Kusema, a Windows dictation app. The file is an unmodified copy, so its size and SHA-256 match what Kusema checks after every download. Kusema's authors did not create or train this model and are not affiliated with or endorsed by its authors. Use the original source below unless you need this copy.
Source and attribution
- Original model: Whisper, by OpenAI (Radford et al., "Robust Speech Recognition via Large-Scale Weak Supervision").
- License: MIT. Copyright OpenAI; the GGML format and tools are copyright the ggml
authors. Both notices are in
LICENSE. - Changes from the original: converted to the GGML format of
whisper.cpp, and some quantized (
q4_1,q5_0). Copied from the files Handy distributed, which Kusema is based on.
Files
| File | Bytes | SHA-256 |
|---|---|---|
ggml-small.bin |
487,601,967 | 1be3a9b2063867b937e64e2ec7483364a79917e157fa98c5d94b5c1fffea987b |
whisper-medium-q4_1.bin |
491,852,915 | 79283fc1f9fe12ca3248543fbd54b73292164d8df5a16e095e2bceeaaabddf57 |
ggml-large-v3-turbo.bin |
1,624,555,275 | 1fc70f774d38eb169993ac391eea357ef47c88757ef72ee5943879b7e8e2bc69 |
ggml-large-v3-q5_0.bin |
1,081,140,203 | d75795ecff3f83b5faa89d1900604ad8c780abd5739fae406de19f23ecd98ad1 |