From: Daniel Bevenius <redacted>
Date: Thu, 3 Apr 2025 17:50:47 +0000 (+0200)
Subject: whisper.wasm : fix unknown language issue (#3000)
X-Git-Tag: upstream/1.7.5+105~99
X-Git-Url: https://git.djapps.eu/?a=commitdiff_plain;h=77e0c86ab62eda9392a8567f4c29ab8d140cb0ba;p=pkg%2Fggml%2Fsources%2Fwhisper.cpp

whisper.wasm : fix unknown language issue (#3000)

* whisper.wasm : fix unknown language issue

This commit addresses an issue with whisper.wasm where the following
error was being displayed when running the application in github pages:
```
whisper_lang_id: unknown language 'д=␙c'
```

This turned out to be a memory corruption issue and further details
can be found in the reference issue below.

Refs: https://github.com/ggerganov/whisper.cpp/issues/2998
---

diff --git a/examples/whisper.wasm/emscripten.cpp b/examples/whisper.wasm/emscripten.cpp
index b84893de..03bf4132 100644
--- a/examples/whisper.wasm/emscripten.cpp
+++ b/examples/whisper.wasm/emscripten.cpp
@@ -65,13 +65,14 @@ EMSCRIPTEN_BINDINGS(whisper) {
         }
 
         struct whisper_full_params params = whisper_full_default_params(whisper_sampling_strategy::WHISPER_SAMPLING_GREEDY);
+        bool is_multilingual = whisper_is_multilingual(g_contexts[index]);
 
         params.print_realtime   = true;
         params.print_progress   = false;
         params.print_timestamps = true;
         params.print_special    = false;
         params.translate        = translate;
-        params.language         = whisper_is_multilingual(g_contexts[index]) ? lang.c_str() : "en";
+        params.language         = is_multilingual ? strdup(lang.c_str()) : "en";
         params.n_threads        = std::min(nthreads, std::min(16, mpow2(std::thread::hardware_concurrency())));
         params.offset_ms        = 0;
 
@@ -102,10 +103,13 @@ EMSCRIPTEN_BINDINGS(whisper) {
 
         // run the worker
         {
-            g_worker = std::thread([index, params, pcmf32 = std::move(pcmf32)]() {
+            g_worker = std::thread([index, params, pcmf32 = std::move(pcmf32), is_multilingual]() {
                 whisper_reset_timings(g_contexts[index]);
                 whisper_full(g_contexts[index], params, pcmf32.data(), pcmf32.size());
                 whisper_print_timings(g_contexts[index]);
+                if (is_multilingual) {
+                    free((void*)params.language);
+                }
             });
         }