Overview

Maintenance release, latest ggml, minor improvements in the tools/server/bindings.

What's Changed

• server : set no_context == true by @ggerganov in https://github.com/ggml-org/whisper.cpp/pull/3482
• WASM example broken for Hebrew by @Orel-A in https://github.com/ggml-org/whisper.cpp/pull/3487
• ruby : tiny bug fix by @KitaitiMakoto in https://github.com/ggml-org/whisper.c...

Overview

• Fix a bug in the ggml norm CPU scalar operator

What's Changed

• sync : ggml by @ggerganov in https://github.com/ggml-org/whisper.cpp/pull/3478

Full Changelog: https://github.com/ggml-org/whisper.cpp/compare/v1.8.1...v1.8.2

Overview

• Fix Vulkan builds
• Fix memory leaks when using VAD
• Support --carry-initial-prompt

What's Changed

• bindings-java : disable flash attention by default by @danbev in https://github.com/ggml-org/whisper.cpp/pull/3445
• ruby : Loose RegExp of a test to follow change of flash attension default by @KitaitiMakoto in https://github.com/ggml-org/whisper.cpp/pull/3448
• vad...

Overview

• Flash attention is now enabled by default
• Performance improvements

M1 Pro

| CPU | Config | Model | Th | FA | Enc. | Dec. | Bch5 | PP | Commit | | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | | M1 Pro | METAL | tiny | 1 | 0 | 32.44 | 1.71 | 0.43 | 0.04 | 8a67c55c |...

Overview

• Add initial VAD support - feedback welcome and appreciated
• Metal FA improvements

M2 Ultra

Flash Attention ON:

| CPU | Config | Model | Th | FA | Enc. | Dec. | Bch5 | PP | Commit | | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | | M2 ULTRA | METAL | tiny | 1 | 1 |...