V1.5 (#121)

v1.5 Former-commit-id: 85560f8
Picovoice · Nov 14, 2018 · 8124a42 · 8124a42
1 parent 9a7f25a
commit 8124a42
Show file tree

Hide file tree

Showing 340 changed files with 373 additions and 300 deletions.
diff --git a/README.md b/README.md
@@ -1,19 +1,20 @@
-# Porcupine &nbsp; [![Tweet](https://img.shields.io/twitter/url/http/shields.io.svg?style=social)](https://twitter.com/intent/tweet?text=On-device%20wake%20word%20detection%20engine%20powered%20by%20deep%20learning&url=https://github.com/Picovoice/Porcupine&hashtags=deeplearning,voice,AI,privacy)
+# Porcupine
 
 [![Build Status](https://travis-ci.org/Picovoice/Porcupine.svg?branch=master)](https://travis-ci.org/Picovoice/Porcupine)
 [![GitHub release](https://img.shields.io/github/release/Picovoice/Porcupine.svg)](https://github.com/Picovoice/Porcupine/releases)
-[![License](https://img.shields.io/badge/License-Apache%202.0-blue.svg)](https://github.com/Picovoice/Porcupine/blob/master/LICENSE)
 
 Made in Vancouver, Canada by [Picovoice](https://picovoice.ai)
 
-Porcupine is a self-service, highly-accurate, and lightweight **wake word** (**voice control**) engine. It enables developers to
-build always-listening voice-enabled applications/platforms. Porcupine is
+Porcupine is a self-service, highly-accurate, and lightweight **wake word** (**voice control**) engine. It enables
+developers to build always-listening voice-enabled applications/platforms. Porcupine is
 
 * self-service. Developers are empowered to **choose any wake word** and build its model **within seconds**.
 * using **deep neural networks** trained in **real-world situations**.
-* compact and computationally-efficient making it suitable for **IoT** applications. It can run with as low as 200 KB of memory.
-* **cross-platform**. It is implemented in fixed-point ANSI C. Currently **Raspberry Pi**, **Android**, **iOS**, **watchOS**, **Linux**,
-**Mac**, **Windows**, and Web browsers are supported. Furthermore, Support for various ARM Cortex-A/M processors and a growing number of DSPs is available.
+* compact and computationally-efficient making it suitable for **IoT** applications. It can run with as low as 200 KB of
+memory.
+* **cross-platform**. It is implemented in fixed-point ANSI C. Currently **Raspberry Pi**, **Android**, **iOS**,
+**watchOS**, **Linux**, **Mac**, **Windows**, and **web browsers** are supported. Furthermore, Support for various
+**ARM Cortex-A** and **ARM Cortex-M** processors and a growing number of **DSPs** is available.
 * **scalable**. It can detect tens of commands concurrently with no added CPU/memory footprint.
 
 ## Table of Contents
@@ -43,8 +44,9 @@ build always-listening voice-enabled applications/platforms. Porcupine is
 
 Try out Porcupine using its [interactive web demo](https://picovoice.ai/#voice-control-demo). You need a working microphone.
 
-Try out Porcupine by downloading it's [Android demo application](https://play.google.com/store/apps/details?id=ai.picovoice.porcupine.demo&hl=en).
-The demo application allows you to test Porcupine on a variety of wake words in any environment.
+Try out Porcupine by downloading it's
+[Android demo application](https://play.google.com/store/apps/details?id=ai.picovoice.porcupine.demo&hl=en). The demo
+application allows you to test Porcupine on a variety of wake words in any environment.
 
 ![Android Demo](resources/images/demo.gif)
 
@@ -60,9 +62,10 @@ The demo application allows you to test Porcupine on a variety of wake words in
 
 ## Performance
 
-A comparison between accuracy and runtime metrics of Porcupine and two other widely-used libraries, PocketSphinx and Snowboy, is provided
-[here](https://github.com/Picovoice/wakeword-benchmark). Compared to the best-performing engine, Porcupine's standard model is
-2.53 times more accurate, 2.6 times faster (on Raspberry Pi 3), and consumes 45% less memory. 
+A comparison between accuracy and runtime metrics of Porcupine and two other widely-used libraries, PocketSphinx and
+Snowboy, is provided [here](https://github.com/Picovoice/wakeword-benchmark). Compared to the best-performing engine,
+Porcupine's standard model is 2.53 times more accurate, 2.6 times faster (on Raspberry Pi 3), and consumes 45% less
+memory. 
 
 ## Model Variants
 
@@ -75,16 +78,18 @@ resources. Below is the comparison of runtime measurements for different variant
 Standard | 7.39% | 1380 KB |
 Tiny | 3.42% | 240 KB |
 
-For accuracy comparison of different variants refer to [benchmark repository](https://github.com/Picovoice/wakeword-benchmark).
+For accuracy comparison of different variants refer to
+[benchmark repository](https://github.com/Picovoice/wakeword-benchmark).
 
 ## Structure of Repository
 
 Porcupine is shipped as an ANSI C precompiled library. The binary files for supported platforms are located under
-[lib/](/lib) and header files are at [include/](/include). Currently, Raspberry Pi, Android, iOS, watchOS, Linux, Mac, and Windows are supported.
-If you wish to run Porcupine on any other platform, please contact us at contact@picovoice.ai.
+[lib/](/lib) and header files are at [include/](/include). Currently, Raspberry Pi, Android, iOS, watchOS, Linux, Mac,
+Windows, and modern web browsers are supported.
 
 Bindings are available at [binding/](/binding) to facilitate usage from higher-level languages/platforms. Demo
-applications are at [demo/](/demo). When possible, use one of the demo applications as a starting point for your own implementation.
+applications are at [demo/](/demo). When possible, use one of the demo applications as a starting point for your own
+implementation.
 
 [tools/](/tools) contains utility programs. Finally, [resources/](/resources) is a placeholder for data used by
 various applications within the repository.
@@ -97,9 +102,9 @@ documentation, it is assumed that the current working directory is the root of t
 ### Python Demo Application
 
 This [demo application](/demo/python) allows testing Porcupine using computer's microphone. It opens an input audio
-stream, monitors it using Porcupine's library, and logs the detection events into the console. Below is an
-example of running the demo for hotword **Alexa** from the command line. Replace `${SYSTEM}` with the name of
-the operating system on your machine (e.g. linux, mac, windows, or raspberry-pi).
+stream, monitors it using Porcupine's library, and logs the detection events into the console. Below is an example of
+running the demo for hotword **Alexa** from the command line. Replace `${SYSTEM}` with the name of the operating system
+on your machine (e.g. linux, mac, windows, or raspberry-pi).
 
 ```bash
 python demo/python/porcupine_demo.py --keyword_file_paths resources/keyword_files/alexa_${SYSTEM}.ppn
@@ -465,6 +470,12 @@ If you like to contribute to Porcupine, please read through [CONTRIBUTING.md](CO
 
 ## Releases
 
+### v1.5.0 - November 13, 2018
+
+* Improved optimizer's accuracy.
+* Runtime optimization.
+* Added support for running within web browsers (WebAssembly).
+
 ### v1.4.0 - July 20, 2018
 
 * Improved accuracy across all models (specifically tiny variant).
@@ -503,5 +514,5 @@ commercial products without acquiring a commercial licensing agreement from Pico
 
 Custom wake-words for other platforms are only provided with the purchase of the commercial license.
 
-In order to inquire
-about the commercial license send an email to contact@picovoice.ai with a brief description of your use case.
+In order to inquire about the commercial license send an email to contact@picovoice.ai with a brief description of your
+use case.
diff --git a/binding/js/porcupine.js b/binding/js/porcupine.js
@@ -11,14 +11,15 @@ let Porcupine = (function () {
     let frameLength;
     let version;
 
-    Module.onRuntimeInitialized = async _ => {
+    let porcupineModule = PorcupineModule();
+    porcupineModule.then(function(Module) {
         initWasm = Module.cwrap('pv_porcupine_wasm_init', 'number', ['number', 'number', 'number', 'number']);
         releaseWasm = Module.cwrap('pv_porcupine_wasm_delete', ['number']);
         processWasm = Module.cwrap('pv_porcupine_wasm_process', 'number', ['number', 'number']);
         sampleRate = Module.cwrap('pv_wasm_sample_rate', 'number', [])();
         frameLength = Module.cwrap('pv_porcupine_wasm_frame_length', 'number', [])();
         version = Module.cwrap('pv_porcupine_wasm_version', 'string', [])();
-    };
+    });
 
     let create = function (keywordIDs, sensitivities) {
         /**
@@ -30,26 +31,26 @@ let Porcupine = (function () {
          * @returns An instance of wake word detection engine.
          */
         let keywordIDSizes = Int32Array.from(keywordIDs.map(keywordID => keywordID.byteLength));
-        let keywordIDSizesPointer = Module._malloc(keywordIDSizes.byteLength);
-        let keywordIDSizesBuffer = new Uint8Array(Module.HEAPU8.buffer, keywordIDSizesPointer, keywordIDSizes.byteLength);
+        let keywordIDSizesPointer = porcupineModule._malloc(keywordIDSizes.byteLength);
+        let keywordIDSizesBuffer = new Uint8Array(porcupineModule.HEAPU8.buffer, keywordIDSizesPointer, keywordIDSizes.byteLength);
         keywordIDSizesBuffer.set(new Uint8Array(keywordIDSizes.buffer));
 
         let keywordIDPointers = Uint32Array.from(keywordIDs.map(keywordID => {
-            let heapPointer = Module._malloc(keywordID.byteLength);
-            let heapBuffer = new Uint8Array(Module.HEAPU8.buffer, heapPointer, keywordID.byteLength);
+            let heapPointer = porcupineModule._malloc(keywordID.byteLength);
+            let heapBuffer = new Uint8Array(porcupineModule.HEAPU8.buffer, heapPointer, keywordID.byteLength);
             heapBuffer.set(keywordID);
             return heapPointer;
         }));
 
-        let keywordIDPointersPointer = Module._malloc(keywordIDPointers.byteLength);
+        let keywordIDPointersPointer = porcupineModule._malloc(keywordIDPointers.byteLength);
         let keywordIDPointersBuffer = new Uint8Array(
-            Module.HEAPU8.buffer,
+            porcupineModule.HEAPU8.buffer,
             keywordIDPointersPointer,
             keywordIDPointers.byteLength);
         keywordIDPointersBuffer.set(new Uint8Array(keywordIDPointers.buffer));
 
-        let sensitivitiesPointer = Module._malloc(sensitivities.byteLength);
-        let sensitivitiesBuffer = new Uint8Array(Module.HEAPU8.buffer, sensitivitiesPointer, sensitivities.byteLength);
+        let sensitivitiesPointer = porcupineModule._malloc(sensitivities.byteLength);
+        let sensitivitiesBuffer = new Uint8Array(porcupineModule.HEAPU8.buffer, sensitivitiesPointer, sensitivities.byteLength);
         sensitivitiesBuffer.set(new Uint8Array(sensitivities.buffer));
 
         let handleWasm = initWasm(
@@ -61,15 +62,15 @@ let Porcupine = (function () {
             throw new Error("failed to initialize porcupine.");
         }
 
-        let pcmWasmPointer = Module._malloc(this.frameLength * 2);
+        let pcmWasmPointer = porcupineModule._malloc(this.frameLength * 2);
 
         let release = function () {
             /**
              * Releases resources acquired by WebAssembly module.
              */
 
             releaseWasm(handleWasm);
-            Module._free(pcmWasmPointer);
+            porcupineModule._free(pcmWasmPointer);
         };
 
         let process = function (pcmInt16Array) {
@@ -80,7 +81,7 @@ let Porcupine = (function () {
              * @returns {Number} Index of detected keyword (phrase). When no keyword is detected it returns -1.
              */
 
-            let pcmWasmBuffer = new Uint8Array(Module.HEAPU8.buffer, pcmWasmPointer, pcmInt16Array.byteLength);
+            let pcmWasmBuffer = new Uint8Array(porcupineModule.HEAPU8.buffer, pcmWasmPointer, pcmInt16Array.byteLength);
             pcmWasmBuffer.set(new Uint8Array(pcmInt16Array.buffer));
 
             let keyword_index = processWasm(handleWasm, pcmWasmPointer);

diff --git a/demo/android/.idea/misc.xml b/demo/android/.idea/misc.xml
diff --git a/demo/android/build.gradle b/demo/android/build.gradle
@@ -22,7 +22,7 @@ buildscript {
         jcenter()
     }
     dependencies {
-        classpath 'com.android.tools.build:gradle:3.1.3'
+        classpath 'com.android.tools.build:gradle:3.1.4'
     }
 }
 

diff --git a/demo/ios/PorcupineDemoNoWatch.xcodeproj/project.pbxproj b/demo/ios/PorcupineDemoNoWatch.xcodeproj/project.pbxproj
@@ -48,26 +48,26 @@
 
 /* Begin PBXFileReference section */
 		BB247359205F636700D96613 /* porcupine_params.pv */ = {isa = PBXFileReference; lastKnownFileType = file; name = porcupine_params.pv; path = ../../../lib/common/porcupine_params.pv; sourceTree = "<group>"; };
-		BB2473622061678900D96613 /* americano_ios.ppn */ = {isa = PBXFileReference; lastKnownFileType = file; path = americano_ios.ppn; sourceTree = "<group>"; };
+		BB2473622061678900D96613 /* americano_ios.ppn */ = {isa = PBXFileReference; lastKnownFileType = text; path = americano_ios.ppn; sourceTree = "<group>"; };
 		BB2473642061679900D96613 /* avocado_ios.ppn */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = text; path = avocado_ios.ppn; sourceTree = "<group>"; };
 		BB247366206167AC00D96613 /* blueberry_ios.ppn */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = text; path = blueberry_ios.ppn; sourceTree = "<group>"; };
-		BB247368206167BA00D96613 /* bumblebee_ios.ppn */ = {isa = PBXFileReference; lastKnownFileType = file; path = bumblebee_ios.ppn; sourceTree = "<group>"; };
+		BB247368206167BA00D96613 /* bumblebee_ios.ppn */ = {isa = PBXFileReference; lastKnownFileType = text; path = bumblebee_ios.ppn; sourceTree = "<group>"; };
 		BB24736A206167C500D96613 /* caterpillar_ios.ppn */ = {isa = PBXFileReference; lastKnownFileType = file; path = caterpillar_ios.ppn; sourceTree = "<group>"; };
 		BB24736C2061687C00D96613 /* christina_ios.ppn */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = text; path = christina_ios.ppn; sourceTree = "<group>"; };
 		BB24736E2061688D00D96613 /* dragonfly_ios.ppn */ = {isa = PBXFileReference; lastKnownFileType = file; path = dragonfly_ios.ppn; sourceTree = "<group>"; };
 		BB2473702061689C00D96613 /* flamingo_ios.ppn */ = {isa = PBXFileReference; lastKnownFileType = file; path = flamingo_ios.ppn; sourceTree = "<group>"; };
 		BB247372206168AB00D96613 /* francesca_ios.ppn */ = {isa = PBXFileReference; lastKnownFileType = text; path = francesca_ios.ppn; sourceTree = "<group>"; };
 		BB247374206168B800D96613 /* grapefruit_ios.ppn */ = {isa = PBXFileReference; lastKnownFileType = file; path = grapefruit_ios.ppn; sourceTree = "<group>"; };
 		BB247376206168C900D96613 /* grasshopper_ios.ppn */ = {isa = PBXFileReference; lastKnownFileType = text; path = grasshopper_ios.ppn; sourceTree = "<group>"; };
-		BB247378206168DA00D96613 /* iguana_ios.ppn */ = {isa = PBXFileReference; lastKnownFileType = file; path = iguana_ios.ppn; sourceTree = "<group>"; };
+		BB247378206168DA00D96613 /* iguana_ios.ppn */ = {isa = PBXFileReference; lastKnownFileType = text; path = iguana_ios.ppn; sourceTree = "<group>"; };
 		BB24737A206168E900D96613 /* picovoice_ios.ppn */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = text; path = picovoice_ios.ppn; sourceTree = "<group>"; };
 		BB24737C2061692800D96613 /* porcupine_ios.ppn */ = {isa = PBXFileReference; lastKnownFileType = file; path = porcupine_ios.ppn; sourceTree = "<group>"; };
 		BB24737E2061693600D96613 /* pineapple_ios.ppn */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = text; path = pineapple_ios.ppn; sourceTree = "<group>"; };
-		BB2473802061694500D96613 /* raspberry_ios.ppn */ = {isa = PBXFileReference; lastKnownFileType = file; path = raspberry_ios.ppn; sourceTree = "<group>"; };
+		BB2473802061694500D96613 /* raspberry_ios.ppn */ = {isa = PBXFileReference; lastKnownFileType = text; path = raspberry_ios.ppn; sourceTree = "<group>"; };
 		BB2473822061694F00D96613 /* vancouver_ios.ppn */ = {isa = PBXFileReference; lastKnownFileType = file; path = vancouver_ios.ppn; sourceTree = "<group>"; };
 		BB2473842061695A00D96613 /* terminator_ios.ppn */ = {isa = PBXFileReference; lastKnownFileType = text; path = terminator_ios.ppn; sourceTree = "<group>"; };
 		BB24738C206312DC00D96613 /* PorcupineManager.swift */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.swift; name = PorcupineManager.swift; path = ../../../binding/ios/PorcupineManager.swift; sourceTree = "<group>"; };
-		BB24738E20633DED00D96613 /* alexa_ios.ppn */ = {isa = PBXFileReference; lastKnownFileType = file; path = alexa_ios.ppn; sourceTree = "<group>"; };
+		BB24738E20633DED00D96613 /* alexa_ios.ppn */ = {isa = PBXFileReference; lastKnownFileType = text; path = alexa_ios.ppn; sourceTree = "<group>"; };
 		BB65C7AF205EEBB300E8770E /* module.modulemap */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = "sourcecode.module-map"; path = module.modulemap; sourceTree = "<group>"; };
 		BB65C7B1205F4BF400E8770E /* libpv_porcupine.a */ = {isa = PBXFileReference; lastKnownFileType = archive.ar; name = libpv_porcupine.a; path = ../../lib/ios/libpv_porcupine.a; sourceTree = "<group>"; };
 		BBE66BCA205DD6F7006005B1 /* PorcupineDemo.app */ = {isa = PBXFileReference; explicitFileType = wrapper.application; includeInIndex = 0; path = PorcupineDemo.app; sourceTree = BUILT_PRODUCTS_DIR; };
@@ -460,6 +460,7 @@
 				CODE_SIGN_STYLE = Automatic;
 				DEVELOPMENT_TEAM = 5KNTYR3JP3;
 				INFOPLIST_FILE = "$(SRCROOT)/PorcupineDemo/Info.plist";
+				IPHONEOS_DEPLOYMENT_TARGET = 12.1;
 				LD_RUNPATH_SEARCH_PATHS = "$(inherited) @executable_path/Frameworks";
 				LIBRARY_SEARCH_PATHS = "${SRCROOT}/../../lib/ios";
 				PRODUCT_BUNDLE_IDENTIFIER = picovoice.porcupine;
@@ -476,6 +477,7 @@
 				CODE_SIGN_STYLE = Automatic;
 				DEVELOPMENT_TEAM = 5KNTYR3JP3;
 				INFOPLIST_FILE = "$(SRCROOT)/PorcupineDemo/Info.plist";
+				IPHONEOS_DEPLOYMENT_TARGET = 12.1;
 				LD_RUNPATH_SEARCH_PATHS = "$(inherited) @executable_path/Frameworks";
 				LIBRARY_SEARCH_PATHS = "${SRCROOT}/../../lib/ios";
 				PRODUCT_BUNDLE_IDENTIFIER = picovoice.porcupine;

diff --git a/demo/js/pv_porcupine.js b/demo/js/pv_porcupine.js
diff --git a/demo/js/pv_porcupine.wasm b/demo/js/pv_porcupine.wasm