Sync with whisper.cpp 🚀

src/whisper/bindings/javascript/package.json

@@ -1,6 +1,6 @@
 {
   "name": "whisper.cpp",
-  "version": "1.5.0",
+  "version": "1.5.1",
   "description": "Whisper speech recognition",
   "main": "whisper.js",
   "scripts": {

src/whisper/examples/server/server.cpp

@@ -72,6 +72,7 @@ struct whisper_params {
     bool no_fallback     = false;
     bool print_special   = false;
     bool print_colors    = false;
+    bool print_realtime  = false;
     bool print_progress  = false;
     bool no_timestamps   = false;
     bool use_gpu         = true;
@@ -144,6 +145,7 @@ void whisper_print_usage(int /*argc*/, char ** argv, const whisper_params & para
     fprintf(stderr, "  -nf,       --no-fallback       [%-7s] do not use temperature fallback while decoding\n", params.no_fallback ? "true" : "false");
     fprintf(stderr, "  -ps,       --print-special     [%-7s] print special tokens\n",                           params.print_special ? "true" : "false");
     fprintf(stderr, "  -pc,       --print-colors      [%-7s] print colors\n",                                   params.print_colors ? "true" : "false");
+    fprintf(stderr, "  -pr,       --print-realtime    [%-7s] print output in realtime\n",                       params.print_realtime ? "true" : "false");
     fprintf(stderr, "  -pp,       --print-progress    [%-7s] print progress\n",                                 params.print_progress ? "true" : "false");
     fprintf(stderr, "  -nt,       --no-timestamps     [%-7s] do not print timestamps\n",                        params.no_timestamps ? "true" : "false");
     fprintf(stderr, "  -l LANG,   --language LANG     [%-7s] spoken language ('auto' for auto-detect)\n",       params.language.c_str());
@@ -188,6 +190,7 @@ bool whisper_params_parse(int argc, char ** argv, whisper_params & params, serve
         else if (arg == "-fp"   || arg == "--font-path")       { params.font_path       = argv[++i]; }
         else if (arg == "-ps"   || arg == "--print-special")   { params.print_special   = true; }
         else if (arg == "-pc"   || arg == "--print-colors")    { params.print_colors    = true; }
+        else if (arg == "-pr"   || arg == "--print-realtime")  { params.print_realtime  = true; }
         else if (arg == "-pp"   || arg == "--print-progress")  { params.print_progress  = true; }
         else if (arg == "-nt"   || arg == "--no-timestamps")   { params.no_timestamps   = true; }
         else if (arg == "-l"    || arg == "--language")        { params.language        = argv[++i]; }
@@ -544,7 +547,7 @@ int main(int argc, char ** argv) {
             whisper_print_user_data user_data = { &params, &pcmf32s, 0 };
 
             // this callback is called on each new segment
-            if (!wparams.print_realtime) {
+            if (params.print_realtime) {
                 wparams.new_segment_callback           = whisper_print_segment_callback;
                 wparams.new_segment_callback_user_data = &user_data;
             }

src/whisper/ggml-metal.h

@@ -52,6 +52,11 @@ void ggml_metal_free(struct ggml_metal_context * ctx);
 void * ggml_metal_host_malloc(size_t n);
 void   ggml_metal_host_free  (void * data);
 
+// helper to check if the device supports a specific family
+// ideally, the user code should be doing these checks
+// ref: https://developer.apple.com/metal/Metal-Feature-Set-Tables.pdf
+bool ggml_metal_supports_family(struct ggml_metal_context * ctx, int family);
+
 // set the number of command buffers to use
 void ggml_metal_set_n_cb(struct ggml_metal_context * ctx, int n_cb);
 
@@ -100,6 +105,8 @@ GGML_API bool ggml_backend_is_metal(ggml_backend_t backend);
 
 GGML_API void ggml_backend_metal_set_n_cb(ggml_backend_t backend, int n_cb);
 
+GGML_API bool ggml_backend_metal_supports_family(ggml_backend_t backend, int family);
+
 #ifdef __cplusplus
 }
 #endif

src/whisper/ggml-metal.m

@@ -459,6 +459,10 @@ void ggml_metal_host_free(void * data) {
     free(data);
 }
 
+bool ggml_metal_supports_family(struct ggml_metal_context * ctx, int family) {
+    return [ctx->device supportsFamily:(MTLGPUFamilyApple1 + family - 1)];
+}
+
 void ggml_metal_set_n_cb(struct ggml_metal_context * ctx, int n_cb) {
     ctx->n_cb = MIN(n_cb, GGML_METAL_MAX_BUFFERS);
 }
@@ -1751,3 +1755,9 @@ void ggml_backend_metal_set_n_cb(ggml_backend_t backend, int n_cb) {
 
     ggml_metal_set_n_cb(ctx, n_cb);
 }
+
+bool ggml_backend_metal_supports_family(ggml_backend_t backend, int family) {
+    struct ggml_metal_context * ctx = (struct ggml_metal_context *)backend->context;
+
+    return ggml_metal_supports_family(ctx, family);
+}

src/whisper/whisper.cpp

@@ -1078,6 +1078,11 @@ static ggml_backend_t whisper_backend_init(const whisper_context_params & params
         if (!backend_gpu) {
             WHISPER_LOG_ERROR("%s: ggml_backend_metal_init() failed\n", __func__);
         }
+        if (!ggml_backend_metal_supports_family(backend_gpu, 7)) {
+            WHISPER_LOG_ERROR("%s: Metal GPU does not support family 7 - falling back to CPU\n", __func__);
+            ggml_backend_free(backend_gpu);
+            backend_gpu = NULL;
+        }
     }
 #endif
 
@@ -3593,6 +3598,17 @@ const char * whisper_lang_str(int id) {
     return nullptr;
 }
 
+const char * whisper_lang_str_full(int id) {
+   for (const auto & kv : g_lang) {
+        if (kv.second.first == id) {
+            return kv.second.second.c_str();
+        }
+    }
+
+    WHISPER_LOG_ERROR("%s: unknown language id %d\n", __func__, id);
+    return nullptr;
+}
+
 int whisper_lang_auto_detect_with_state(
         struct whisper_context * ctx,
           struct whisper_state * state,
@@ -5180,7 +5196,7 @@ int whisper_full_with_state(
                 ctx, state, progress_cur, params.progress_callback_user_data);
         }
 
-        // of only 1 second left, then stop
+        // if only 1 second left, then stop
         if (seek + 100 >= seek_end) {
             break;
         }

src/whisper/whisper.h

@@ -51,7 +51,7 @@ extern "C" {
     //     ...
     //
     //     whisper_context_params cparams = whisper_context_default_params();
-    // 
+    //
     //     struct whisper_context * ctx = whisper_init_from_file_with_params("/path/to/ggml-base.en.bin", cparams);
     //
     //     if (whisper_full(ctx, wparams, pcmf32.data(), pcmf32.size()) != 0) {
@@ -315,6 +315,9 @@ extern "C" {
     // Return the short string of the specified language id (e.g. 2 -> "de"), returns nullptr if not found
     WHISPER_API const char * whisper_lang_str(int id);
 
+    // Return the short string of the specified language name (e.g. 2 -> "german"), returns nullptr if not found
+    WHISPER_API const char * whisper_lang_str_full(int id);
+
     // Use mel data at offset_ms to try and auto-detect the spoken language
     // Make sure to call whisper_pcm_to_mel() or whisper_set_mel() first
     // Returns the top language id or negative on failure