V1.0 (#36)

README.md

@@ -23,9 +23,6 @@ voice assistants. Orca is:
     - Raspberry Pi (3, 4, 5)
     - Chrome, Safari, Firefox, and Edge
 
-**Please note that Orca is currently in development. While we prioritize stability and compatibility, certain aspects of
-Orca may undergo changes as we continually enhance and refine the engine to provide the best user experience possible.**
-
 ## Table of Contents
 
 - [Orca](#orca)
@@ -57,15 +54,15 @@ Orca may undergo changes as we continually enhance and refine the engine to prov
 
 ## Language Support
 
-- Orca Text-to-Speech currently supports English only.
+- Orca Streaming Text-to-Speech currently supports English only.
 - Support for [additional languages is available for commercial customers](https://picovoice.ai/consulting/) on a
   case-by-case basis.
 
 ## Overview
 
 ### Orca input and output streaming synthesis
 
-Orca is a text-to-speech engine designed specifically for LLMs. It can process
+Orca is a streaming text-to-speech engine designed specifically for LLMs. It can process
 incoming text streams in real-time, generating audio continuously, i.e., as the LLM produces tokens,
 Orca generates speech in parallel.
 This enables seamless conversations with voice assistants, eliminating any audio delays.
@@ -76,9 +73,9 @@ Orca also supports single synthesis mode, where a complete text is synthesized i
 
 ### Text input
 
-Orca accepts the 26 lowercase (a-z) and 26 uppercase (A-Z) letters of the English alphabet, numbers,
-basic symbols, as well as common punctuation marks. You can get a list of all supported characters by calling the
-`valid_characters()` method provided in the Orca SDK you are using.
+Orca supports a wide range of English characters, including letters, numbers, symbols, and punctuation marks. 
+You can get a list of all supported characters by calling the `valid_characters()` method provided 
+in the Orca SDK you are using.
 Pronunciations of characters or words not supported by this list can be achieved with
 [custom pronunciations](#custom-pronunciations).
 
@@ -683,6 +680,13 @@ For more details, see the [Node.js SDK](./binding/nodejs/).
 
 ## Releases
 
+### v1.0.0 - Aug 20th, 2024
+
+- Improved voice quality
+- Significantly reduced latency in streaming synthesis
+- Reduced model size
+- Advanced text normalization
+
 ### v0.2.0 - May 3rd, 2024
 
 - Support for streaming synthesis

binding/android/Orca/orca/build.gradle

@@ -2,7 +2,7 @@ apply plugin: 'com.android.library'
 
 ext {
     PUBLISH_GROUP_ID = 'ai.picovoice'
-    PUBLISH_VERSION = '0.2.1'
+    PUBLISH_VERSION = '1.0.0'
     PUBLISH_ARTIFACT_ID = 'orca-android'
 }
 
@@ -15,7 +15,7 @@ android {
         minSdkVersion 21
         targetSdkVersion defaultTargetSdkVersion
         versionCode 2
-        versionName "0.2.1"
+        versionName "1.0.0"
 
         consumerProguardFiles "consumer-rules.pro"
     }

binding/android/OrcaTestApp/orca-test-app/build.gradle

@@ -106,7 +106,7 @@ dependencies {
     implementation 'androidx.appcompat:appcompat:1.6.1'
     implementation 'com.google.android.material:material:1.8.0'
     implementation 'androidx.constraintlayout:constraintlayout:2.1.4'
-    implementation 'ai.picovoice:orca-android:0.2.1'
+    implementation 'ai.picovoice:orca-android:1.0.0'
 
     // Espresso UI Testing
     androidTestImplementation 'androidx.test.ext:junit:1.1.5'
@@ -115,7 +115,7 @@ dependencies {
     })
     androidTestImplementation('com.microsoft.appcenter:espresso-test-extension:1.4')
     androidTestImplementation('androidx.test.espresso:espresso-intents:3.5.1')
-    androidTestImplementation('ai.picovoice:orca-android:0.2.1')
+    androidTestImplementation('ai.picovoice:orca-android:1.0.0')
 }
 
 afterEvaluate {

binding/android/OrcaTestApp/orca-test-app/src/androidTest/java/ai/picovoice/orca/testapp/OrcaTest.java

@@ -12,7 +12,6 @@
 
 package ai.picovoice.orca.testapp;
 
-import static org.junit.Assert.assertArrayEquals;
 import static org.junit.Assert.assertEquals;
 import static org.junit.Assert.assertNotEquals;
 import static org.junit.Assert.assertNotNull;

binding/ios/Orca-iOS.podspec

@@ -1,7 +1,7 @@
 Pod::Spec.new do |s|
     s.name = 'Orca-iOS'
     s.module_name = 'Orca'
-    s.version = '0.2.2'
+    s.version = '1.0.0'
     s.license = {:type => 'Apache 2.0'}
     s.summary = 'iOS binding for Picovoice\'s Orca Text-to-Speech Engine.'
     s.description =
@@ -18,7 +18,7 @@ Pod::Spec.new do |s|
     DESC
     s.homepage = 'https://github.com/Picovoice/orca/tree/main/binding/ios'
     s.author = { 'Picovoice' => 'hello@picovoice.ai' }
-    s.source = { :git => "https://github.com/Picovoice/orca.git", :tag => "Orca-iOS-v0.2.2" }
+    s.source = { :git => "https://github.com/Picovoice/orca.git", :tag => "Orca-iOS-v1.0.0" }
     s.ios.deployment_target = '13.0'
     s.swift_version = '5.0'
     s.vendored_frameworks = 'lib/ios/PvOrca.xcframework'

binding/ios/OrcaAppTest/OrcaAppTestUITests/BaseTest.swift

@@ -102,7 +102,7 @@ class BaseTest: XCTestCase {
     }
 
     func compareArrays(arr1: [Int16], arr2: [Int16], step: Int) -> Bool {
-        for i in stride(from: 0, to: arr1.count - step, by: step) where !(abs(arr1[i] - arr2[i]) <= 500) {
+        for i in stride(from: 0, to: arr1.count - step, by: step) where !(abs(arr1[i] - arr2[i]) <= 12000) {
             return false
         }
         return true

binding/ios/OrcaAppTest/Podfile

@@ -2,13 +2,13 @@ source 'https://cdn.cocoapods.org/'
 platform :ios, '13.0'
 
 target 'OrcaAppTest' do
-  pod 'Orca-iOS', '~> 0.2.2'
+  pod 'Orca-iOS', '~> 1.0.0'
 end
 
 target 'OrcaAppTestUITests' do
-  pod 'Orca-iOS', '~> 0.2.2'
+  pod 'Orca-iOS', '~> 1.0.0'
 end
 
 target 'PerformanceTest' do
-  pod 'Orca-iOS', '~> 0.2.2'
+  pod 'Orca-iOS', '~> 1.0.0'
 end

binding/ios/OrcaAppTest/Podfile.lock

@@ -1,16 +1,16 @@
 PODS:
-  - Orca-iOS (0.2.2)
+  - Orca-iOS (1.0.0)
 
 DEPENDENCIES:
-  - Orca-iOS (~> 0.2.2)
+  - Orca-iOS (~> 1.0.0)
 
 SPEC REPOS:
   trunk:
     - Orca-iOS
 
 SPEC CHECKSUMS:
-  Orca-iOS: 567ca0e53671d8fc28ba15338db8fe4ff5101d8d
+  Orca-iOS: d50a0dbbf596f20c6c2e2f727f20f72ac012aa0e
 
-PODFILE CHECKSUM: 2deb98490df78cf895d797180dcfeccea4f2ad25
+PODFILE CHECKSUM: 85110c27ad76f2840ec98aff4c162e9ffc71a022
 
 COCOAPODS: 1.15.2

binding/nodejs/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@picovoice/orca-node",
-  "version": "0.2.1",
+  "version": "1.0.0",
   "description": "Picovoice Orca Node.js binding",
   "main": "dist/index.js",
   "types": "dist/types/index.d.ts",

binding/nodejs/src/platforms.ts

@@ -31,7 +31,6 @@ const PLATFORM_WINDOWS = 'windows';
 
 const ARM_CPU_64 = '-aarch64';
 const ARM_CPU_CORTEX_A53 = 'cortex-a53';
-const ARM_CPU_CORTEX_A57 = 'cortex-a57';
 const ARM_CPU_CORTEX_A72 = 'cortex-a72';
 const ARM_CPU_CORTEX_A76 = 'cortex-a76';
 
@@ -121,8 +120,6 @@ function getLinuxMachine(arch: string): string {
   switch (cpuPart) {
     case '0xd03':
       return ARM_CPU_CORTEX_A53 + archInfo;
-    case '0xd07':
-      return ARM_CPU_CORTEX_A57 + archInfo;
     case '0xd08':
       return ARM_CPU_CORTEX_A72 + archInfo;
     case '0xd0b':

binding/python/_util.py

@@ -37,8 +37,6 @@ def _linux_machine() -> str:
 
     if "0xd03" == cpu_part:
         return "cortex-a53" + arch_info
-    elif "0xd07" == cpu_part:
-        return "cortex-a57" + arch_info
     elif "0xd08" == cpu_part:
         return "cortex-a72" + arch_info
     elif "0xd0b" == cpu_part:

binding/python/setup.py

@@ -49,7 +49,7 @@
 
 setuptools.setup(
     name="pvorca",
-    version="0.2.4",
+    version="1.0.0",
     author="Picovoice",
     author_email="hello@picovoice.ai",
     description="Orca Streaming Text-to-Speech Engine",

binding/python/test_orca.py

@@ -23,8 +23,6 @@
 
 
 class OrcaTestCase(unittest.TestCase):
-    EXACT_ALIGNMENT_TEST_MODEL_IDENTIFIER = "female"
-
     access_key: str
     orcas: List[Orca]
     model_paths: List[str]
@@ -48,7 +46,7 @@ def _test_audio(self, pcm: Sequence[int], ground_truth: Sequence[int]) -> None:
         pcm = pcm[:len(ground_truth)]  # compensate for discrepancies due to wav header
         self.assertEqual(len(pcm), len(ground_truth))
         for i in range(len(pcm)):
-            self.assertAlmostEqual(pcm[i], ground_truth[i], delta=8000)
+            self.assertAlmostEqual(pcm[i], ground_truth[i], delta=12000)
 
     def _test_equal_timestamp(self, timestamp: float, timestamp_truth: float) -> None:
         self.assertAlmostEqual(timestamp, timestamp_truth, places=2)
@@ -105,7 +103,7 @@ def test_synthesize(self) -> None:
     def test_synthesize_alignment_exact(self) -> None:
         orca = [
             orca for i, orca in enumerate(self.orcas) if
-            self.EXACT_ALIGNMENT_TEST_MODEL_IDENTIFIER in self.model_paths[i]].pop()
+            test_data.exact_alignment_test_model_identifier in self.model_paths[i]].pop()
         pcm, alignments = orca.synthesize(test_data.text_alignment, random_state=test_data.random_state)
         self.assertGreater(len(pcm), 0)
 
@@ -115,7 +113,7 @@ def test_synthesize_alignment_exact(self) -> None:
 
     def test_synthesize_alignment(self) -> None:
         for i, orca in enumerate(self.orcas):
-            if self.EXACT_ALIGNMENT_TEST_MODEL_IDENTIFIER in self.model_paths[i]:
+            if test_data.exact_alignment_test_model_identifier in self.model_paths[i]:
                 continue
 
             pcm, alignments = orca.synthesize(test_data.text_alignment, random_state=test_data.random_state)

binding/python/test_util.py

@@ -29,6 +29,7 @@ class TestData:
     alignments: Sequence[Orca.WordAlignment]
     random_state: int
     audio_data_folder: str
+    exact_alignment_test_model_identifier: str
 
 
 def read_wav_file(path: str) -> Sequence[int]:
@@ -49,7 +50,7 @@ def get_test_data() -> TestData:
         test_data = json.loads(data_file.read())
 
     alignments = []
-    for word_data in test_data["alignments"]:
+    for word_data in test_data.pop("alignments"):
         phonemes = []
         for phoneme_data in word_data["phonemes"]:
             phoneme = Orca.PhonemeAlignment(
@@ -65,11 +66,11 @@ def get_test_data() -> TestData:
             phonemes=phonemes)
         alignments.append(word)
 
+    test_sentences = test_data.pop("test_sentences")
     test_data = TestData(
         alignments=alignments,
-        random_state=test_data["random_state"],
-        audio_data_folder=test_data["audio_data_folder"],
-        **test_data["test_sentences"])
+        **test_data,
+        **test_sentences)
 
     return test_data
 

binding/web/package.json

@@ -3,7 +3,7 @@
   "description": "Orca Text-to-Speech engine for web browsers (via WebAssembly)",
   "author": "Picovoice Inc",
   "license": "Apache-2.0",
-  "version": "0.2.1",
+  "version": "1.0.0",
   "keywords": [
     "orca",
     "web",
@@ -33,7 +33,7 @@
     "test-perf": "cypress run --spec test/orca_perf.test.ts"
   },
   "dependencies": {
-    "@picovoice/web-utils": "=1.3.4"
+    "@picovoice/web-utils": "=1.4.2"
   },
   "devDependencies": {
     "@babel/core": "^7.21.3",

binding/web/src/orca.ts

@@ -589,8 +589,8 @@ export class Orca {
             throw new OrcaErrors.OrcaOutOfMemoryError('malloc failed: Cannot allocate memory');
           }
 
-          const memoryBufferView = new DataView(this._wasmMemory.buffer);
-          const memoryBufferUint8 = new Uint8Array(this._wasmMemory.buffer);
+          let memoryBufferView = new DataView(this._wasmMemory.buffer);
+          let memoryBufferUint8 = new Uint8Array(this._wasmMemory.buffer);
 
           const initStatus = await this._pvOrcaSynthesizeParamsInit(synthesizeParamsAddressAddress);
           if (initStatus !== PvStatus.SUCCESS) {
@@ -690,6 +690,9 @@ export class Orca {
           await this._pvFree(textAddress);
           await this._pvOrcaSynthesizeParamsDelete(synthesizeParamsAddress);
 
+          memoryBufferView = new DataView(this._wasmMemory.buffer);
+          memoryBufferUint8 = new Uint8Array(this._wasmMemory.buffer);
+
           if (synthesizeStatus !== PvStatus.SUCCESS) {
             const messageStack = await Orca.getMessageStack(
               this._pvGetErrorStack,
@@ -928,7 +931,7 @@ export class Orca {
 
   private static async initWasm(accessKey: string, modelPath: string, wasmBase64: string): Promise<OrcaWasmOutput> {
     // A WebAssembly page has a constant size of 64KiB. -> 1MiB ~= 16 pages
-    const memory = new WebAssembly.Memory({ initial: 7500 });
+    const memory = new WebAssembly.Memory({ initial: 1600 });
     const memoryBufferUint8 = new Uint8Array(memory.buffer);
     const pvError = new PvError();
     const exports = await buildWasm(memory, wasmBase64, pvError);
@@ -1025,6 +1028,7 @@ export class Orca {
       objectAddressAddress);
     await pv_free(accessKeyAddress);
     await pv_free(modelPathAddress);
+
     if (initStatus !== PvStatus.SUCCESS) {
       const messageStack = await Orca.getMessageStack(
         pv_get_error_stack,

binding/web/src/orca_worker_handler.ts

@@ -92,7 +92,7 @@ self.onmessage = async function(
           self.postMessage({
             command: 'error',
             status: PvStatus.RUNTIME_ERROR,
-            shortMessage: 'Orca synthesize error',
+            shortMessage: e.message,
           });
         }
       }
@@ -133,7 +133,7 @@ self.onmessage = async function(
           self.postMessage({
             command: 'error',
             status: PvStatus.RUNTIME_ERROR,
-            shortMessage: 'Orca stream open error',
+            shortMessage: e.message,
           });
         }
       }
@@ -172,7 +172,7 @@ self.onmessage = async function(
           self.postMessage({
             command: 'error',
             status: PvStatus.RUNTIME_ERROR,
-            shortMessage: 'Orca synthesize error',
+            shortMessage: e.message,
           });
         }
       }

binding/web/test/orca.test.ts

@@ -29,11 +29,11 @@ const EXPECTED_VALID_CHARACTERS = [
   '-', '/', '1', '2', '3', '4', '5',
   '6', '7', '8', '9', '0', '@', '%',
   '&', '\n', '_', '(', ')', '°', 'º',
-  '²', '³',
+  '²', '³', '$', '€', '¥', '₪', '£',
+  '₩', '₺', '₱', '₽', '฿', '₴', '₹',
+  '¢', '+', '=',
 ];
 
-const EXACT_ALIGNMENT_TEST_MODEL_IDENTIFIER = 'female';
-
 const compareArrays = (arr1: Int16Array, arr2: Int16Array, step: number) => {
   expect(arr1.length).eq(arr2.length);
   for (let i = 0; i < arr1.length - step; i += step) {
@@ -179,7 +179,7 @@ describe('Orca Binding', function() {
         }
       });
 
-      if (modelFileSuffix === EXACT_ALIGNMENT_TEST_MODEL_IDENTIFIER) {
+      if (publicPath.includes(testData.exact_alignment_test_model_identifier)) {
         it(`should be able to process alignment exact [${modelFileSuffix}] (${instanceString})`, async () => {
           try {
             const orca = await instance.create(

binding/web/yarn.lock

@@ -1122,10 +1122,10 @@
   dependencies:
     commander "^9.2.0"
 
-"@picovoice/web-utils@=1.3.4":
-  version "1.3.4"
-  resolved "https://registry.yarnpkg.com/@picovoice/web-utils/-/web-utils-1.3.4.tgz#de3c4afe39aaf6e620320e413bfdf69bd9921f5d"
-  integrity sha512-f5bh1jvlbK53EzGdbpbcvNDoz16wbuQ/Fkj+T0GDQulowky+u/HfGZJE7si4vQ/m/2Lctx07X8YxmUHlH2FKGg==
+"@picovoice/web-utils@=1.4.2":
+  version "1.4.2"
+  resolved "https://registry.yarnpkg.com/@picovoice/web-utils/-/web-utils-1.4.2.tgz#2ddc44552d15fa1a4958e0c3384e58545255eea1"
+  integrity sha512-pF5Uw3Vm4mOWJ2H3Zc7E/nDr/O7OhbvgEK6W7cx9MNNK3qq51MqiGluPpZ8a2K61BuIzxcNMC1mXWpmIAWVolA==
   dependencies:
     commander "^10.0.1"