Merge remote-tracking branch 'ggerganov/master'

* ggerganov/master: (72 commits)
  ci : use selective copy for musa image (ggml-org#3296)
  ci: set fail-fast to false in docker.yml (ggml-org#3294)
  ruby : add Whisper::VERSION (ggml-org#3292)
  whisper : add version function (ggml-org#3289)
  ci : add should_release variable (ggml-org#3288)
  docs : add cmake "-j" flag in README.md (ggml-org#3284)
  ci : add support for tag-based releases (ggml-org#3287)
  release : v1.7.6
  bench : update benches
  bench : print system info before ctx check
  stream : add nullptr check of whisper_context (ggml-org#3283)
  ci : enable main-cuda build (ggml-org#3282)
  bindings.java : update java example (ggml-org#3281)
  coreml : backport CoreML features to macos < 14 (ggml-org#3255)
  ci : reduce musa image size (ggml-org#3277)
  whisper : add .gitignore entries for OpenVINO support (ggml-org#3276)
  command: output commands to text file (ggml-org#3273)
  ci : add apt-get clean to musa Dockerfile (ggml-org#3275)
  ruby : specify Apple frameworks explicitly on build (ggml-org#3270)
  talk-llama : sync llama.cpp
  ...

Author: bygreencn

.devops/main-cuda.Dockerfile

@@ -16,6 +16,7 @@ ENV CUDA_DOCKER_ARCH=${CUDA_DOCKER_ARCH}
 
 RUN apt-get update && \
     apt-get install -y build-essential libsdl2-dev wget cmake git \
+    && apt-get clean \
     && rm -rf /var/lib/apt/lists/* /var/cache/apt/archives/*
 
 # Ref: https://stackoverflow.com/a/53464012
@@ -26,15 +27,24 @@ COPY .. .
 # Enable cuBLAS
 RUN make base.en CMAKE_ARGS="-DGGML_CUDA=1"
 
+RUN find /app/build -name "*.o" -delete && \
+    find /app/build -name "*.a" -delete && \
+    rm -rf /app/build/CMakeFiles && \
+    rm -rf /app/build/cmake_install.cmake && \
+    rm -rf /app/build/_deps
+
 FROM ${BASE_CUDA_RUN_CONTAINER} AS runtime
 ENV CUDA_MAIN_VERSION=12.3
 ENV LD_LIBRARY_PATH /usr/local/cuda-${CUDA_MAIN_VERSION}/compat:$LD_LIBRARY_PATH
 WORKDIR /app
 
 RUN apt-get update && \
   apt-get install -y curl ffmpeg wget cmake git \
+  && apt-get clean \
   && rm -rf /var/lib/apt/lists/* /var/cache/apt/archives/*
 
 COPY --from=build /app /app
+RUN du -sh /app/*
+RUN find /app -type f -size +100M
 ENV PATH=/app/build/bin:$PATH
 ENTRYPOINT [ "bash", "-c" ]

.devops/main-musa.Dockerfile

@@ -10,20 +10,31 @@ FROM ${BASE_MUSA_DEV_CONTAINER} AS build
 WORKDIR /app
 
 RUN apt-get update && \
-    apt-get install -y build-essential libsdl2-dev wget cmake git \
-    && rm -rf /var/lib/apt/lists/* /var/cache/apt/archives/*
+    apt-get install -y build-essential libsdl2-dev wget cmake git && \
+    apt-get clean && \
+    rm -rf /var/lib/apt/lists/* /var/cache/apt/archives/* /tmp/* /var/tmp/*
 
 COPY .. .
 # Enable muBLAS
 RUN make base.en CMAKE_ARGS="-DGGML_MUSA=1"
 
+RUN find /app/build -name "*.o" -delete && \
+    find /app/build -name "*.a" -delete && \
+    rm -rf /app/build/CMakeFiles && \
+    rm -rf /app/build/cmake_install.cmake && \
+    rm -rf /app/build/_deps
+
 FROM ${BASE_MUSA_RUN_CONTAINER} AS runtime
 WORKDIR /app
 
 RUN apt-get update && \
-  apt-get install -y curl ffmpeg wget cmake git \
-  && rm -rf /var/lib/apt/lists/* /var/cache/apt/archives/*
+    apt-get install -y curl ffmpeg wget cmake git && \
+    apt-get clean && \
+    rm -rf /var/lib/apt/lists/* /var/cache/apt/archives/* /tmp/* /var/tmp/*
+
+COPY --from=build /app/build/bin /app/build/bin
+COPY --from=build /app/samples /app/samples
+COPY --from=build /app/models /app/models
 
-COPY --from=build /app /app
 ENV PATH=/app/build/bin:$PATH
 ENTRYPOINT [ "bash", "-c" ]

.github/workflows/build.yml

@@ -4,6 +4,8 @@ on:
   push:
     branches:
       - master
+    tags:
+      - 'v*'
   pull_request:
     types: [opened, synchronize, reopened]
   workflow_dispatch:
@@ -41,6 +43,7 @@ jobs:
     runs-on: ubuntu-latest
     outputs:
       tag_name: ${{ steps.tag.outputs.name }}
+      should_release: ${{ steps.tag.outputs.should_release }}
 
     steps:
       - name: Checkout with full history
@@ -55,28 +58,42 @@ jobs:
           BUILD_NUMBER=$(git rev-list --count HEAD)
           SHORT_HASH=$(git rev-parse --short=7 HEAD)
           CUSTOM_TAG="${{ github.event.inputs.pre_release_tag }}"
+          SHOULD_RELEASE="false"
 
           echo "Raw values:"
           echo "BUILD_NUMBER: $BUILD_NUMBER"
           echo "SHORT_HASH: $SHORT_HASH"
           echo "BRANCH_NAME: ${{ env.BRANCH_NAME }}"
           echo "CUSTOM_TAG: $CUSTOM_TAG"
 
-          # Use custom tag if provided
-          if [[ -n "$CUSTOM_TAG" ]]; then
+          if [[ "${{ github.ref_type }}" == "tag" ]]; then
+            echo "Using pushed tag name"
+            TAG_NAME="${{ github.ref_name }}"
+            SHOULD_RELEASE="true"
+          elif [[ -n "$CUSTOM_TAG" ]]; then
             echo "Using custom tag"
             TAG_NAME="${CUSTOM_TAG}"
+            SHOULD_RELEASE="true"
+          elif [[ "${{ github.event.inputs.create_release }}" == "true" ]]; then
+            echo "Manual release requested"
+            SHOULD_RELEASE="true"
+            TAG_NAME="b${BUILD_NUMBER}"
           elif [[ "${{ env.BRANCH_NAME }}" == "master" ]]; then
             echo "Using master branch format"
             TAG_NAME="b${BUILD_NUMBER}"
+            SHOULD_RELEASE="false"
           else
             echo "Using non-master branch format"
             SAFE_NAME=$(echo "${{ env.BRANCH_NAME }}" | tr '/' '-')
             TAG_NAME="${SAFE_NAME}-b${BUILD_NUMBER}-${SHORT_HASH}"
+            SHOULD_RELEASE="false"
           fi
 
           echo "Final tag name: $TAG_NAME"
+          echo "Should release: $SHOULD_RELEASE"
           echo "name=$TAG_NAME" >> $GITHUB_OUTPUT
+          echo "should_release=$SHOULD_RELEASE" >> $GITHUB_OUTPUT
+
 
   ubuntu-22:
     if: ${{ github.event_name == 'push' || github.event_name == 'pull_request' ||
@@ -579,6 +596,7 @@ jobs:
     if: ${{ github.event_name == 'push' || github.event_name == 'pull_request' ||
             github.event.inputs.run_type == 'full-ci' }}
     runs-on: windows-latest
+    needs: determine-tag
 
     strategy:
       matrix:
@@ -662,9 +680,7 @@ jobs:
               Compress-Archive -Path "build/bin/${{ matrix.build }}" -DestinationPath "whisper-bin-${{ matrix.arch }}.zip"
 
       - name: Upload binaries
-        if: matrix.sdl2 == 'ON' && ${{ (github.event_name == 'push' && github.ref == 'refs/heads/master') ||
-                github.event.inputs.create_release == 'true' ||
-                github.event.inputs.pre_release_tag != '' }}
+        if: matrix.sdl2 == 'ON' && ${{ needs.determine-tag.outputs.should_release }}
         uses: actions/upload-artifact@v4
         with:
           name: whisper-bin-${{ matrix.arch }}.zip
@@ -750,9 +766,7 @@ jobs:
               Compress-Archive -Path "build/bin/${{ matrix.build }}" -DestinationPath "whisper-blas-bin-${{ matrix.arch }}.zip"
 
       - name: Upload binaries
-        if: matrix.blas == 'ON' && matrix.sdl2 == 'ON' && ${{ (github.event_name == 'push' && github.ref == 'refs/heads/master') ||
-                github.event.inputs.create_release == 'true' ||
-                github.event.inputs.pre_release_tag != '' }}
+        if: matrix.blas == 'ON' && matrix.sdl2 == 'ON' && ${{ needs.determine-tag.outputs.should_release }}
         uses: actions/upload-artifact@v4
         with:
           name: whisper-blas-bin-${{ matrix.arch }}.zip
@@ -762,6 +776,7 @@ jobs:
     if: ${{ github.event_name == 'push' || github.event_name == 'pull_request' ||
             github.event.inputs.run_type == 'full-ci' }}
     runs-on: windows-2022
+    needs: determine-tag
     strategy:
       fail-fast: false
       matrix:
@@ -960,9 +975,7 @@ jobs:
               Compress-Archive -Path "build/bin/${{ matrix.build }}" -DestinationPath "whisper-cublas-${{ matrix.cuda-toolkit }}-bin-${{ matrix.arch }}.zip"
 
       - name: Upload binaries
-        if: ${{ (github.event_name == 'push' && github.ref == 'refs/heads/master') ||
-                github.event.inputs.create_release == 'true' ||
-                github.event.inputs.pre_release_tag != '' }}
+        if: ${{ needs.determine-tag.outputs.should_release }}
         uses: actions/upload-artifact@v4
         with:
           name: whisper-cublas-${{ matrix.cuda-toolkit }}-bin-${{ matrix.arch }}.zip
@@ -1039,16 +1052,11 @@ jobs:
 
       - name: Pack artifacts
         id: pack_artifacts
-        if: ${{ (github.event_name == 'push' && github.ref == 'refs/heads/master') ||
-                github.event.inputs.create_release == 'true' ||
-                github.event.inputs.pre_release_tag != '' }}
         run: |
           zip --symlinks -r whisper-${{ needs.determine-tag.outputs.tag_name }}-xcframework.zip build-apple/whisper.xcframework
 
       - name: Upload artifacts
-        if: ${{ (github.event_name == 'push' && github.ref == 'refs/heads/master') ||
-                github.event.inputs.create_release == 'true' ||
-                github.event.inputs.pre_release_tag != '' }}
+        if: ${{ needs.determine-tag.outputs.should_release }}
         uses: actions/upload-artifact@v4
         with:
           path: whisper-${{ needs.determine-tag.outputs.tag_name }}-xcframework.zip
@@ -1226,7 +1234,7 @@ jobs:
           ./build/bin/quantize models/ggml-tiny.en.bin models/ggml-tiny.en-q4_0.bin q4_0
 
   release:
-    if: ${{ github.event.inputs.create_release == 'true' || github.event.inputs.pre_release_tag != '' }}
+    if: ${{ github.event.inputs.create_release == 'true' || github.event.inputs.pre_release_tag != '' || startsWith(github.ref, 'refs/tags/v') }}
 
     runs-on: ubuntu-latest
 
@@ -1269,6 +1277,7 @@ jobs:
         with:
           tag_name: ${{ needs.determine-tag.outputs.tag_name }}
           prerelease: ${{ github.event.inputs.pre_release_tag != '' }}
+          draft: true
 
       - name: Upload release
         id: upload_release
@@ -1295,7 +1304,8 @@ jobs:
   coreml-base-en:
     if: ${{ (github.event_name == 'push' && github.ref == 'refs/heads/master') ||
             github.event.inputs.create_release == 'true' ||
-            github.event.inputs.pre_release_tag != '' }}
+            github.event.inputs.pre_release_tag != '' ||
+            startsWith(github.ref, 'refs/tags/v') }}
     runs-on: macos-latest
     needs: determine-tag
 

.github/workflows/docker.yml

@@ -15,14 +15,13 @@ jobs:
     env:
       COMMIT_SHA: ${{ github.sha }}
     strategy:
+      fail-fast: false
       matrix:
         config:
           - { tag: "main", dockerfile: ".devops/main.Dockerfile", platform: "linux/amd64" }
           - { tag: "main-musa", dockerfile: ".devops/main-musa.Dockerfile", platform: "linux/amd64" }
           - { tag: "main-intel", dockerfile: ".devops/main-intel.Dockerfile", platform: "linux/amd64" }
-          #TODO: the cuda image keeps failing - disable for now
-          #      https://github.com/ggerganov/whisper.cpp/actions/runs/11019444428/job/30602020339
-          #- { tag: "main-cuda", dockerfile: ".devops/main-cuda.Dockerfile", platform: "linux/amd64" }
+          - { tag: "main-cuda", dockerfile: ".devops/main-cuda.Dockerfile", platform: "linux/amd64" }
 
     steps:
       - name: Check out the repo
@@ -43,21 +42,35 @@ jobs:
           username: ${{ github.repository_owner }}
           password: ${{ secrets.GITHUB_TOKEN }}
 
-      - name: Build and push Docker image (versioned)
-        if: github.event_name == 'push'
-        uses: docker/build-push-action@v5
-        with:
-          context: .
-          push: true
-          platforms: ${{ matrix.config.platform }}
-          tags: "ghcr.io/${{ github.repository }}:${{ matrix.config.tag }}-${{ env.COMMIT_SHA }}"
-          file: ${{ matrix.config.dockerfile }}
+      - name: Free up disk space
+        run: |
+          sudo apt-get remove -y '^dotnet-.*' '^llvm-.*' '^mysql-.*' '^postgresql-.*'
+          sudo apt-get autoremove -y
+          sudo apt-get autoclean
+
+          sudo rm -rf /usr/share/dotnet
+          sudo rm -rf /usr/local/lib/android
+          sudo rm -rf /opt/ghc
+          sudo rm -rf /opt/hostedtoolcache/CodeQL
+
+          docker system prune -af
+
+          df -h
+
+      - name: Generate tags
+        id: tags
+        run: |
+          TAGS="ghcr.io/${{ github.repository }}:${{ matrix.config.tag }}"
+          if [ "${{ github.event_name }}" == "push" ]; then
+            TAGS="$TAGS,ghcr.io/${{ github.repository }}:${{ matrix.config.tag }}-${{ env.COMMIT_SHA }}"
+          fi
+          echo "tags=$TAGS" >> $GITHUB_OUTPUT
 
       - name: Build and push Docker image (tagged)
-        uses: docker/build-push-action@v4
+        uses: docker/build-push-action@v5
         with:
           context: .
           push: ${{ github.event_name == 'push' }}
           platforms: ${{ matrix.config.platform }}
-          tags: "ghcr.io/${{ github.repository }}:${{ matrix.config.tag }}"
+          tags: ${{ steps.tags.outputs.tags }}
           file: ${{ matrix.config.dockerfile }}

.gitignore

@@ -50,6 +50,8 @@ extra/bench-gg.txt
 models/*.mlmodel
 models/*.mlmodelc
 models/*.mlpackage
+models/*-encoder-openvino.xml
+models/*-encoder-openvino-cache/
 bindings/java/.gradle/
 bindings/java/.idea/
 .idea/

CMakeLists.txt

@@ -1,6 +1,6 @@
 cmake_minimum_required(VERSION 3.5) # for add_link_options and implicit target directories.
 project("whisper.cpp" C CXX)
-project("whisper.cpp" VERSION 1.7.5)
+project("whisper.cpp" VERSION 1.7.6)
 include(CheckIncludeFileCXX)
 
 set(SOVERSION 1)
@@ -182,6 +182,10 @@ get_directory_property(WHISPER_TRANSIENT_DEFINES COMPILE_DEFINITIONS)
 set_target_properties(whisper PROPERTIES PUBLIC_HEADER ${CMAKE_CURRENT_SOURCE_DIR}/include/whisper.h)
 install(TARGETS whisper LIBRARY PUBLIC_HEADER)
 
+target_compile_definitions(whisper PRIVATE
+    WHISPER_VERSION="${PROJECT_VERSION}"
+)
+
 configure_package_config_file(
         ${CMAKE_CURRENT_SOURCE_DIR}/cmake/whisper-config.cmake.in
         ${CMAKE_CURRENT_BINARY_DIR}/whisper-config.cmake

README.md

@@ -7,7 +7,7 @@
 [![Conan Center](https://shields.io/conan/v/whisper-cpp)](https://conan.io/center/whisper-cpp)
 [![npm](https://img.shields.io/npm/v/whisper.cpp.svg)](https://www.npmjs.com/package/whisper.cpp/)
 
-Stable: [v1.7.5](https://github.com/ggml-org/whisper.cpp/releases/tag/v1.7.5) / [Roadmap](https://github.com/orgs/ggml-org/projects/4/)
+Stable: [v1.7.6](https://github.com/ggml-org/whisper.cpp/releases/tag/v1.7.6) / [Roadmap](https://github.com/orgs/ggml-org/projects/4/)
 
 High-performance inference of [OpenAI's Whisper](https://github.com/openai/whisper) automatic speech recognition (ASR) model:
 
@@ -80,7 +80,7 @@ Now build the [whisper-cli](examples/cli) example and transcribe an audio file l
 ```bash
 # build the project
 cmake -B build
-cmake --build build --config Release
+cmake --build build -j --config Release
 
 # transcribe an audio file
 ./build/bin/whisper-cli -f samples/jfk.wav
@@ -149,7 +149,7 @@ standard cmake setup with:
 ```bash
 # build with GGML_BLAS defined
 cmake -B build -DGGML_BLAS=1
-cmake --build build --config Release
+cmake --build build -j --config Release
 ./build/bin/whisper-cli [ .. etc .. ]
 ```
 
@@ -163,7 +163,7 @@ Here are the steps for creating and using a quantized model:
 ```bash
 # quantize a model with Q5_0 method
 cmake -B build
-cmake --build build --config Release
+cmake --build build -j --config Release
 ./build/bin/quantize models/ggml-base.en.bin models/ggml-base.en-q5_0.bin q5_0
 
 # run the examples as usual, specifying the quantized model file
@@ -489,7 +489,7 @@ You will need to have [sdl2](https://wiki.libsdl.org/SDL2/Installation) installe
 
 ```bash
 cmake -B build -DWHISPER_SDL2=ON
-cmake --build build --config Release
+cmake --build build -j --config Release
 ./build/bin/whisper-stream -m ./models/ggml-base.en.bin -t 8 --step 500 --length 5000
 ```
 
@@ -709,7 +709,9 @@ For more details, see the conversion script [models/convert-pt-to-ggml.py](model
 ## XCFramework
 The XCFramework is a precompiled version of the library for iOS, visionOS, tvOS,
 and macOS. It can be used in Swift projects without the need to compile the
-library from source. For examples:
+library from source. For example, the v1.7.5 version of the XCFramework can be
+used as follows:
+
 ```swift
 // swift-tools-version: 5.10
 // The swift-tools-version declares the minimum version of Swift required to build this package.