release : v1.4.3

* whisper : add support for large v3

* bench : fix build + fix go bindings

* bench : fix n_mels

* models : update readme

CMakeLists.txt

@@ -1,6 +1,6 @@
 cmake_minimum_required (VERSION 3.5)
 
-project(whisper.cpp VERSION 1.4.2)
+project(whisper.cpp VERSION 1.4.3)
 
 # Add path to modules
 list(APPEND CMAKE_MODULE_PATH "${CMAKE_CURRENT_SOURCE_DIR}/cmake/")

README.md

@@ -6,7 +6,7 @@
 [![License: MIT](https://img.shields.io/badge/license-MIT-blue.svg)](https://opensource.org/licenses/MIT)
 [![npm](https://img.shields.io/npm/v/whisper.cpp.svg)](https://www.npmjs.com/package/whisper.cpp/)
 
-Beta: [v1.4.2](https://github.com/ggerganov/whisper.cpp/releases/tag/v1.4.2) / Stable: [v1.2.1](https://github.com/ggerganov/whisper.cpp/releases/tag/v1.2.1) / [Roadmap | F.A.Q.](https://github.com/ggerganov/whisper.cpp/discussions/126)
+Beta: [v1.4.3](https://github.com/ggerganov/whisper.cpp/releases/tag/v1.4.3) / Stable: [v1.2.1](https://github.com/ggerganov/whisper.cpp/releases/tag/v1.2.1) / [Roadmap | F.A.Q.](https://github.com/ggerganov/whisper.cpp/discussions/126)
 
 High-performance inference of [OpenAI's Whisper](https://github.com/openai/whisper) automatic speech recognition (ASR) model:
 

bindings/javascript/package.json

@@ -1,6 +1,6 @@
 {
   "name": "whisper.cpp",
-  "version": "1.4.2",
+  "version": "1.4.3",
   "description": "Whisper speech recognition",
   "main": "whisper.js",
   "scripts": {

examples/whisper.android/app/build.gradle

@@ -18,9 +18,7 @@ android {
         vectorDrawables {
             useSupportLibrary true
         }
-        ndk {
-            abiFilters 'arm64-v8a', 'armeabi-v7a', 'x86', 'x86_64'
-        }
+
     }
 
     buildTypes {
@@ -43,20 +41,10 @@ android {
     composeOptions {
         kotlinCompilerExtensionVersion '1.5.0'
     }
-    ndkVersion "25.2.9519653"
-    externalNativeBuild {
-        cmake {
-            path = file("src/main/jni/whisper/CMakeLists.txt")
-        }
-    }
-    packagingOptions {
-        resources {
-            excludes += '/META-INF/{AL2.0,LGPL2.1}'
-        }
-    }
 }
 
 dependencies {
+    implementation project(':lib')
     implementation 'androidx.activity:activity-compose:1.7.2'
     implementation 'androidx.compose.material:material-icons-core:1.5.0'
     implementation 'androidx.compose.material3:material3:1.1.1'

examples/whisper.android/app/src/main/java/com/whispercppdemo/ui/main/MainScreenViewModel.kt

@@ -15,7 +15,7 @@ import androidx.lifecycle.viewmodel.initializer
 import androidx.lifecycle.viewmodel.viewModelFactory
 import com.whispercppdemo.media.decodeWaveFile
 import com.whispercppdemo.recorder.Recorder
-import com.whispercppdemo.whisper.WhisperContext
+import com.whispercpp.whisper.WhisperContext
 import kotlinx.coroutines.Dispatchers
 import kotlinx.coroutines.launch
 import kotlinx.coroutines.runBlocking
@@ -35,7 +35,7 @@ class MainScreenViewModel(private val application: Application) : ViewModel() {
     private val modelsPath = File(application.filesDir, "models")
     private val samplesPath = File(application.filesDir, "samples")
     private var recorder: Recorder = Recorder()
-    private var whisperContext: WhisperContext? = null
+    private var whisperContext: com.whispercpp.whisper.WhisperContext? = null
     private var mediaPlayer: MediaPlayer? = null
     private var recordedFile: File? = null
 
@@ -47,7 +47,7 @@ class MainScreenViewModel(private val application: Application) : ViewModel() {
     }
 
     private suspend fun printSystemInfo() {
-        printMessage(String.format("System Info: %s\n", WhisperContext.getSystemInfo()))
+        printMessage(String.format("System Info: %s\n", com.whispercpp.whisper.WhisperContext.getSystemInfo()))
     }
 
     private suspend fun loadData() {
@@ -78,7 +78,7 @@ class MainScreenViewModel(private val application: Application) : ViewModel() {
         printMessage("Loading model...\n")
         val models = application.assets.list("models/")
         if (models != null) {
-            whisperContext = WhisperContext.createContextFromAsset(application.assets, "models/" + models[0])
+            whisperContext = com.whispercpp.whisper.WhisperContext.createContextFromAsset(application.assets, "models/" + models[0])
             printMessage("Loaded model ${models[0]}.\n")
         }
 

examples/whisper.android/lib/.gitignore

@@ -0,0 +1 @@
+/build

examples/whisper.android/lib/build.gradle

@@ -0,0 +1,51 @@
+plugins {
+    id 'com.android.library'
+    id 'org.jetbrains.kotlin.android'
+}
+
+android {
+    namespace 'com.whispercpp'
+    compileSdk 34
+
+    defaultConfig {
+        minSdk 26
+        targetSdk 34
+        versionCode 1
+        versionName "1.0"
+
+        ndk {
+            abiFilters 'arm64-v8a', 'armeabi-v7a', 'x86', 'x86_64'
+        }
+    }
+
+    buildTypes {
+        release {
+            minifyEnabled false
+        }
+    }
+    compileOptions {
+        sourceCompatibility JavaVersion.VERSION_1_8
+        targetCompatibility JavaVersion.VERSION_1_8
+    }
+    kotlinOptions {
+        jvmTarget = '1.8'
+    }
+
+    ndkVersion "25.2.9519653"
+    externalNativeBuild {
+        cmake {
+            path = file("src/main/jni/whisper/CMakeLists.txt")
+        }
+    }
+    packagingOptions {
+        resources {
+            excludes += '/META-INF/{AL2.0,LGPL2.1}'
+        }
+    }
+}
+
+dependencies {
+    implementation 'androidx.core:core-ktx:1.9.0'
+    implementation 'androidx.appcompat:appcompat:1.6.1'
+    implementation 'com.google.android.material:material:1.8.0'
+}

examples/whisper.android/lib/src/main/AndroidManifest.xml

@@ -0,0 +1,4 @@
+<?xml version="1.0" encoding="utf-8"?>
+<manifest xmlns:android="http://schemas.android.com/apk/res/android">
+
+</manifest>

examples/whisper.android/lib/src/main/java/com/whispercpp/whisper/LibWhisper.kt

@@ -1,4 +1,4 @@
-package com.whispercppdemo.whisper
+package com.whispercpp.whisper
 
 import android.content.res.AssetManager
 import android.os.Build

examples/whisper.android/lib/src/main/java/com/whispercpp/whisper/WhisperCpuConfig.kt

@@ -1,4 +1,4 @@
-package com.whispercppdemo.whisper
+package com.whispercpp.whisper
 
 import android.util.Log
 import java.io.BufferedReader

examples/whisper.android/lib/src/main/jni/whisper/jni.c

@@ -131,7 +131,7 @@ static struct whisper_context *whisper_init_from_asset(
 }
 
 JNIEXPORT jlong JNICALL
-Java_com_whispercppdemo_whisper_WhisperLib_00024Companion_initContextFromAsset(
+Java_com_whispercpp_whisper_WhisperLib_00024Companion_initContextFromAsset(
         JNIEnv *env, jobject thiz, jobject assetManager, jstring asset_path_str) {
     UNUSED(thiz);
     struct whisper_context *context = NULL;
@@ -142,7 +142,7 @@ Java_com_whispercppdemo_whisper_WhisperLib_00024Companion_initContextFromAsset(
 }
 
 JNIEXPORT jlong JNICALL
-Java_com_whispercppdemo_whisper_WhisperLib_00024Companion_initContext(
+Java_com_whispercpp_whisper_WhisperLib_00024Companion_initContext(
         JNIEnv *env, jobject thiz, jstring model_path_str) {
     UNUSED(thiz);
     struct whisper_context *context = NULL;
@@ -153,7 +153,7 @@ Java_com_whispercppdemo_whisper_WhisperLib_00024Companion_initContext(
 }
 
 JNIEXPORT void JNICALL
-Java_com_whispercppdemo_whisper_WhisperLib_00024Companion_freeContext(
+Java_com_whispercpp_whisper_WhisperLib_00024Companion_freeContext(
         JNIEnv *env, jobject thiz, jlong context_ptr) {
     UNUSED(env);
     UNUSED(thiz);
@@ -162,7 +162,7 @@ Java_com_whispercppdemo_whisper_WhisperLib_00024Companion_freeContext(
 }
 
 JNIEXPORT void JNICALL
-Java_com_whispercppdemo_whisper_WhisperLib_00024Companion_fullTranscribe(
+Java_com_whispercpp_whisper_WhisperLib_00024Companion_fullTranscribe(
         JNIEnv *env, jobject thiz, jlong context_ptr, jint num_threads, jfloatArray audio_data) {
     UNUSED(thiz);
     struct whisper_context *context = (struct whisper_context *) context_ptr;
@@ -194,7 +194,7 @@ Java_com_whispercppdemo_whisper_WhisperLib_00024Companion_fullTranscribe(
 }
 
 JNIEXPORT jint JNICALL
-Java_com_whispercppdemo_whisper_WhisperLib_00024Companion_getTextSegmentCount(
+Java_com_whispercpp_whisper_WhisperLib_00024Companion_getTextSegmentCount(
         JNIEnv *env, jobject thiz, jlong context_ptr) {
     UNUSED(env);
     UNUSED(thiz);
@@ -203,7 +203,7 @@ Java_com_whispercppdemo_whisper_WhisperLib_00024Companion_getTextSegmentCount(
 }
 
 JNIEXPORT jstring JNICALL
-Java_com_whispercppdemo_whisper_WhisperLib_00024Companion_getTextSegment(
+Java_com_whispercpp_whisper_WhisperLib_00024Companion_getTextSegment(
         JNIEnv *env, jobject thiz, jlong context_ptr, jint index) {
     UNUSED(thiz);
     struct whisper_context *context = (struct whisper_context *) context_ptr;
@@ -213,7 +213,7 @@ Java_com_whispercppdemo_whisper_WhisperLib_00024Companion_getTextSegment(
 }
 
 JNIEXPORT jstring JNICALL
-Java_com_whispercppdemo_whisper_WhisperLib_00024Companion_getSystemInfo(
+Java_com_whispercpp_whisper_WhisperLib_00024Companion_getSystemInfo(
         JNIEnv *env, jobject thiz
 ) {
     UNUSED(thiz);
@@ -223,7 +223,7 @@ Java_com_whispercppdemo_whisper_WhisperLib_00024Companion_getSystemInfo(
 }
 
 JNIEXPORT jstring JNICALL
-Java_com_whispercppdemo_whisper_WhisperLib_00024Companion_benchMemcpy(JNIEnv *env, jobject thiz,
+Java_com_whispercpp_whisper_WhisperLib_00024Companion_benchMemcpy(JNIEnv *env, jobject thiz,
                                                                       jint n_threads) {
     UNUSED(thiz);
     const char *bench_ggml_memcpy = whisper_bench_memcpy_str(n_threads);
@@ -231,7 +231,7 @@ Java_com_whispercppdemo_whisper_WhisperLib_00024Companion_benchMemcpy(JNIEnv *en
 }
 
 JNIEXPORT jstring JNICALL
-Java_com_whispercppdemo_whisper_WhisperLib_00024Companion_benchGgmlMulMat(JNIEnv *env, jobject thiz,
+Java_com_whispercpp_whisper_WhisperLib_00024Companion_benchGgmlMulMat(JNIEnv *env, jobject thiz,
                                                                           jint n_threads) {
     UNUSED(thiz);
     const char *bench_ggml_mul_mat = whisper_bench_ggml_mul_mat_str(n_threads);

examples/whisper.android/settings.gradle

@@ -14,3 +14,4 @@ dependencyResolutionManagement {
 }
 rootProject.name = "WhisperCppDemo"
 include ':app'
+include ':lib'

ggml.c

@@ -143,12 +143,6 @@ void ggml_print_backtrace(void) {
 }
 #endif
 
-#undef MIN
-#undef MAX
-
-#define MIN(a, b) ((a) < (b) ? (a) : (b))
-#define MAX(a, b) ((a) > (b) ? (a) : (b))
-
 /*#define GGML_PERF*/
 #define GGML_DEBUG 0
 #define GGML_GELU_FP16
@@ -277,6 +271,12 @@ inline static void * ggml_aligned_malloc(size_t size) {
 // floating point type used to accumulate sums
 typedef double ggml_float;
 
+#undef MIN
+#undef MAX
+
+#define MIN(a, b) ((a) < (b) ? (a) : (b))
+#define MAX(a, b) ((a) > (b) ? (a) : (b))
+
 //
 // global data
 //

whisper.cpp

@@ -3714,6 +3714,7 @@ void whisper_print_timings(struct whisper_context * ctx) {
 }
 
 void whisper_reset_timings(struct whisper_context * ctx) {
+    ctx->t_start_us = ggml_time_us();
     if (ctx->state != nullptr) {
         ctx->state->t_sample_us = 0;
         ctx->state->t_encode_us = 0;