Add speaker identification

.coveragerc

@@ -3,6 +3,7 @@ omit =
   buzz/whisper_cpp.py
   *_test.py
   demucs/*
+  whisper_diarization/*
 
 [html]
 directory = coverage/html

.github/workflows/ci.yml

@@ -147,7 +147,7 @@ jobs:
 
       - name: Downgrade torch for Intel macOS
         run: |
-          poetry add torch==2.2.2 torchaudio==2.2.2
+          poetry add torch==2.2.2 torchaudio==2.2.2 nemo_toolkit[asr]==2.0.0
         if: "matrix.os == 'macos-13'"
 
       - name: Install dependencies

.github/workflows/snapcraft.yml

@@ -15,9 +15,22 @@ concurrency:
 jobs:
   build:
     runs-on: ubuntu-latest
+    timeout-minutes: 90
+    env:
+      BUZZ_DISABLE_TELEMETRY: true
     outputs:
       snap: ${{ steps.snapcraft.outputs.snap }}
     steps:
+      # Ideas from https://github.com/orgs/community/discussions/25678
+      - name: Remove unused build tools
+        run: |
+          sudo apt-get remove -y '^llvm-.*'
+          sudo apt-get remove -y 'php.*'
+          sudo apt-get remove -y azure-cli google-cloud-sdk hhvm google-chrome-stable firefox powershell mono-devel || true
+          sudo apt-get autoremove -y
+          sudo apt-get clean
+          python -m pip cache purge
+          rm -rf /opt/hostedtoolcache || true
       - name: Maximize build space
         uses: easimon/maximize-build-space@master
         with:
@@ -27,6 +40,7 @@ jobs:
           remove-android: 'true'
           remove-haskell: 'true'
           remove-codeql: 'true'
+          remove-docker-images: 'true'
       - uses: actions/checkout@v4
         with:
           submodules: recursive

.gitignore

@@ -32,4 +32,5 @@ benchmarks.json
 /coverage/
 /wheelhouse/
 /.flatpak-builder
-/repo
+/repo
+/nemo_msdd_configs

.gitmodules

@@ -1,3 +1,6 @@
 [submodule "whisper.cpp"]
 	path = whisper.cpp
 	url = https://github.com/ggerganov/whisper.cpp
+[submodule "whisper_diarization"]
+	path = whisper_diarization
+	url = https://github.com/MahmoudAshraf97/whisper-diarization

Buzz.spec

@@ -28,7 +28,8 @@ datas += copy_metadata("pyyaml")
 datas += collect_data_files("transformers", include_py_files=True)
 
 datas += collect_data_files("whisper")
-datas += [("demucs", "demucs")]
+datas += collect_data_files("demucs", include_py_files=True)
+datas += collect_data_files("whisper_diarization", include_py_files=True)
 datas += [("buzz/assets/*", "assets")]
 datas += [("buzz/locale", "locale")]
 datas += [("buzz/schema.sql", ".")]

Makefile

@@ -45,7 +45,7 @@ else
 	rm -rf dist/* || true
 endif
 
-COVERAGE_THRESHOLD := 75
+COVERAGE_THRESHOLD := 70
 
 test: buzz/whisper_cpp.py translation_mo
 	pytest -s -vv --cov=buzz --cov-report=xml --cov-report=html --benchmark-skip --cov-fail-under=${COVERAGE_THRESHOLD} --cov-config=.coveragerc
@@ -237,7 +237,7 @@ ifeq ($(OS), Windows_NT)
 	done
 else
 	for dir in buzz/locale/*/ ; do \
-		python msgfmt.py -o $$dir/LC_MESSAGES/buzz.mo $$dir/LC_MESSAGES/buzz.po; \
+		python3 msgfmt.py -o $$dir/LC_MESSAGES/buzz.mo $$dir/LC_MESSAGES/buzz.po; \
 	done
 endif