Merge branch 'master' into allow-for-NextSegment-to-be-called-when-Process-is-re-called

Author: joshmux

CMakeLists.txt

@@ -34,6 +34,9 @@ endif()
 if (EMSCRIPTEN)
     set(BUILD_SHARED_LIBS_DEFAULT OFF)
 
+    set(CMAKE_CXX_STANDARD 17)
+    set(CMAKE_CXX_STANDARD_REQUIRED ON)
+
     option(WHISPER_WASM_SINGLE_FILE "whisper: embed WASM inside the generated whisper.js" ON)
 
     # TODO: without these, we get the following error:

models/for-tests-silero-v5.1.2-ggml.bin

@@ -6693,7 +6693,7 @@ static bool whisper_vad(
             }
 
             segment_start_samples = std::min(segment_start_samples, n_samples - 1);
-            segment_end_samples = std::min(segment_end_samples, n_samples);
+            segment_end_samples = std::min(segment_end_samples, n_samples - 1);
             int segment_length = segment_end_samples - segment_start_samples;
             if (segment_length > 0) {
                 whisper_state::vad_segment_info segment;

models/for-tests-silero-v6.2.0-ggml.bin

@@ -94,7 +94,7 @@ add_executable(${VAD_TEST} ${VAD_TEST}.cpp)
 target_include_directories(${VAD_TEST} PRIVATE ../include ../ggml/include ../examples)
 target_link_libraries(${VAD_TEST} PRIVATE common)
 target_compile_definitions(${VAD_TEST} PRIVATE
-    VAD_MODEL_PATH="${PROJECT_SOURCE_DIR}/models/for-tests-silero-v5.1.2-ggml.bin"
+    VAD_MODEL_PATH="${PROJECT_SOURCE_DIR}/models/for-tests-silero-v6.2.0-ggml.bin"
     SAMPLE_PATH="${PROJECT_SOURCE_DIR}/samples/jfk.wav")
 add_test(NAME ${VAD_TEST} COMMAND ${VAD_TEST})
 set_tests_properties(${VAD_TEST} PROPERTIES LABELS "unit")
@@ -106,7 +106,7 @@ target_include_directories(${VAD_TEST} PRIVATE ../include ../ggml/include ../exa
 target_link_libraries(${VAD_TEST} PRIVATE common)
 target_compile_definitions(${VAD_TEST} PRIVATE
     WHISPER_MODEL_PATH="${PROJECT_SOURCE_DIR}/models/ggml-base.en.bin"
-    VAD_MODEL_PATH="${PROJECT_SOURCE_DIR}/models/for-tests-silero-v5.1.2-ggml.bin"
+    VAD_MODEL_PATH="${PROJECT_SOURCE_DIR}/models/for-tests-silero-v6.2.0-ggml.bin"
     SAMPLE_PATH="${PROJECT_SOURCE_DIR}/samples/jfk.wav")
 add_test(NAME ${VAD_TEST} COMMAND ${VAD_TEST})
 set_tests_properties(${VAD_TEST} PROPERTIES LABELS "base;en")

src/whisper.cpp

@@ -42,11 +42,13 @@ int main() {
     const int n_segments = whisper_full_n_segments(wctx);
     assert(n_segments == 1);
 
+
+    printf("Segment text:\n%s", whisper_full_get_segment_text(wctx, 0));
     assert(strcmp(" And so my fellow Americans, ask not what your country can do for you,"
                   " ask what you can do for your country.",
            whisper_full_get_segment_text(wctx, 0)) == 0);
-    assert(whisper_full_get_segment_t0(wctx, 0) == 29);
-    assert(whisper_full_get_segment_t1(wctx, 0) == 1049);
+    assert(whisper_full_get_segment_t0(wctx, 0) == 32);
+    assert(whisper_full_get_segment_t1(wctx, 0) == 1051);
 
     whisper_free(wctx);
 

tests/CMakeLists.txt

@@ -36,7 +36,7 @@ struct whisper_vad_segments * test_detect_timestamps(
         struct whisper_vad_context * vctx,
         struct whisper_vad_params params) {
     struct whisper_vad_segments * timestamps = whisper_vad_segments_from_probs(vctx, params);
-    assert(whisper_vad_segments_n_segments(timestamps) == 5);
+    assert(whisper_vad_segments_n_segments(timestamps) == 4);
 
     for (int i = 0; i < whisper_vad_segments_n_segments(timestamps); ++i) {
         printf("VAD segment %d: start = %.2f, end = %.2f\n", i,