v1.8.5

[ggerganov]

Overview

Maintenance release + performance improvements all around:

https://github.com/ggml-org/whisper.cpp/blob/master/scripts/bench-all-gg.txt

What's Changed

• ruby : fix dangling pointers, memory leak, and SEGV on parallel transcription by @KitaitiMakoto in ruby : fix dangling pointers, memory leak, and SEGV on parallel transcription #3715
• sync : ggml by @ggerganov in sync : ggml #3730
• vad : add streaming detect + explicit state reset by @danielbodart in vad : add streaming detect + explicit state reset #3677
• bench : sync submit-results URL to ggml-org by @jinweihan-ai in bench : sync submit-results URL to ggml-org #3769
• sync : ggml by @ggerganov in sync : ggml #3786
• sync : ggml by @ggerganov in sync : ggml #3789
• ruby : transcribe without GVL, accept more MemoryViews, Windows support, fix memory size report, improve document by @KitaitiMakoto in ruby : transcribe without GVL, accept more MemoryViews, Windows support, fix memory size report, improve document #3775
• Incorrect timestamps by @bviksoe in Incorrect timestamps #2279
• server: Add support for controlling token_timestamps directly by @alubbe in server: Add support for controlling token_timestamps directly #3785
• Fix SetMaxTokensPerSegment in go binding causes the segements to end prematurely #984: About max_tokens skipping remaining audio issue by @annaeina in Fix #984: About max_tokens skipping remaining audio issue #3798
• server: Fix params leak between requests by @alubbe in server: Fix params leak between requests #3784
• server: Fix no_speech_thold not being read by @alubbe in server: Fix no_speech_thold not being read #3783
• sync : ggml by @ggerganov in sync : ggml #3803
• server: Add support for carry_initial_prompt by @alubbe in server: Add support for carry_initial_prompt #3781
• server: Return speaker information in JSON by @alubbe in server: Return speaker information in JSON #3782
• examples : fix memory leak in read_audio_data by @petterreinholdtsen in examples : fix memory leak in read_audio_data #3810
• whisper : set bench data for each iteration by @danbev in whisper : set bench data for each iteration #3812
• ci : use github ubuntu-22.04-arm runner instead of qemu by @danbev in ci : use github ubuntu-22.04-arm runner instead of qemu #3815
• fix: server /inference fails to decode in-memory audio (regression) by @ServeurpersoCom in fix: server /inference fails to decode in-memory audio (regression) #3818
• fix: in bindings/ruby/test/jfk_reader/jfk_reader in jfk_reader.c by @orbisai0security in fix: in bindings/ruby/test/jfk_reader/jfk_reader in jfk_reader.c #3756
• cmake : add CMakePresets.json [no ci] by @danbev in cmake : add CMakePresets.json [no ci] #3808
• sync : ggml by @ggerganov in sync : ggml #3824
• [AMD] Add AMD ROCm/HIP GPU build instructions by @Kaihui-AMD in [AMD] Add AMD ROCm/HIP GPU build instructions #3823
• cli : merge tokens split across UTF-8 boundaries in JSON output by @texasich in cli : merge tokens split across UTF-8 boundaries in JSON output #3751
• docs : add AGENTS.md and CONTRIBUTING.md [no ci] by @danbev in docs : add AGENTS.md and CONTRIBUTING.md [no ci] #3826
• ci : only run docker jobs when pushed to master [no ci] by @danbev in ci : only run docker jobs when pushed to master [no ci] #3828
• ci : set GGML_NATIVE=OFF for bindings-java by @danbev in ci : set GGML_NATIVE=OFF for bindings-java #3830
• ci : renable arm64 docker builds by @danbev in ci : renable arm64 docker builds #3832
• ci : add on push/pull_request paths ruby job by @danbev in ci : add on push/pull_request paths ruby job #3833
• ci : fix include paths for bindings-go job [no ci] by @danbev in ci : fix include paths for bindings-go job [no ci] #3835
• ci : add ignore for bindings/{ruby, go} in build.yml [no ci] by @danbev in ci : add ignore for bindings/{ruby, go} in build.yml [no ci] #3837
• sync : ggml by @ggerganov in sync : ggml #3840

New Contributors

• @danielbodart made their first contribution in vad : add streaming detect + explicit state reset #3677
• @jinweihan-ai made their first contribution in bench : sync submit-results URL to ggml-org #3769
• @bviksoe made their first contribution in Incorrect timestamps #2279
• @annaeina made their first contribution in Fix #984: About max_tokens skipping remaining audio issue #3798
• @ServeurpersoCom made their first contribution in fix: server /inference fails to decode in-memory audio (regression) #3818
• @orbisai0security made their first contribution in fix: in bindings/ruby/test/jfk_reader/jfk_reader in jfk_reader.c #3756
• @Kaihui-AMD made their first contribution in [AMD] Add AMD ROCm/HIP GPU build instructions #3823
• @texasich made their first contribution in cli : merge tokens split across UTF-8 boundaries in JSON output #3751

Full Changelog: v1.8.4...v1.8.5

---

This discussion was created from the release v1.8.5.

[yousef03330777-commits]

Zaxscdvfv b