@capgo/capacitor-speech-recognition 7.0.0

package/dist/plugin.cjs.js

@@ -0,0 +1,39 @@
+'use strict';
+
+var core = require('@capacitor/core');
+
+const SpeechRecognition = core.registerPlugin('SpeechRecognition', {
+    web: () => Promise.resolve().then(function () { return web; }).then((m) => new m.SpeechRecognitionWeb()),
+});
+
+class SpeechRecognitionWeb extends core.WebPlugin {
+    available() {
+        throw this.unimplemented('Speech recognition is not available on the web.');
+    }
+    start(_options) {
+        throw this.unimplemented('Speech recognition is not available on the web.');
+    }
+    stop() {
+        throw this.unimplemented('Speech recognition is not available on the web.');
+    }
+    getSupportedLanguages() {
+        throw this.unimplemented('Speech recognition is not available on the web.');
+    }
+    isListening() {
+        throw this.unimplemented('Speech recognition is not available on the web.');
+    }
+    checkPermissions() {
+        throw this.unimplemented('Speech recognition permissions are not handled on the web.');
+    }
+    requestPermissions() {
+        throw this.unimplemented('Speech recognition permissions are not handled on the web.');
+    }
+}
+
+var web = /*#__PURE__*/Object.freeze({
+    __proto__: null,
+    SpeechRecognitionWeb: SpeechRecognitionWeb
+});
+
+exports.SpeechRecognition = SpeechRecognition;
+//# sourceMappingURL=plugin.cjs.js.map

package/dist/plugin.cjs.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"plugin.cjs.js","sources":["esm/index.js","esm/web.js"],"sourcesContent":["import { registerPlugin } from '@capacitor/core';\nconst SpeechRecognition = registerPlugin('SpeechRecognition', {\n    web: () => import('./web').then((m) => new m.SpeechRecognitionWeb()),\n});\nexport * from './definitions';\nexport { SpeechRecognition };\n//# sourceMappingURL=index.js.map","import { WebPlugin } from '@capacitor/core';\nexport class SpeechRecognitionWeb extends WebPlugin {\n    available() {\n        throw this.unimplemented('Speech recognition is not available on the web.');\n    }\n    start(_options) {\n        throw this.unimplemented('Speech recognition is not available on the web.');\n    }\n    stop() {\n        throw this.unimplemented('Speech recognition is not available on the web.');\n    }\n    getSupportedLanguages() {\n        throw this.unimplemented('Speech recognition is not available on the web.');\n    }\n    isListening() {\n        throw this.unimplemented('Speech recognition is not available on the web.');\n    }\n    checkPermissions() {\n        throw this.unimplemented('Speech recognition permissions are not handled on the web.');\n    }\n    requestPermissions() {\n        throw this.unimplemented('Speech recognition permissions are not handled on the web.');\n    }\n}\n//# sourceMappingURL=web.js.map"],"names":["registerPlugin","WebPlugin"],"mappings":";;;;AACK,MAAC,iBAAiB,GAAGA,mBAAc,CAAC,mBAAmB,EAAE;AAC9D,IAAI,GAAG,EAAE,MAAM,mDAAe,CAAC,IAAI,CAAC,CAAC,CAAC,KAAK,IAAI,CAAC,CAAC,oBAAoB,EAAE,CAAC;AACxE,CAAC;;ACFM,MAAM,oBAAoB,SAASC,cAAS,CAAC;AACpD,IAAI,SAAS,GAAG;AAChB,QAAQ,MAAM,IAAI,CAAC,aAAa,CAAC,iDAAiD,CAAC;AACnF,IAAI;AACJ,IAAI,KAAK,CAAC,QAAQ,EAAE;AACpB,QAAQ,MAAM,IAAI,CAAC,aAAa,CAAC,iDAAiD,CAAC;AACnF,IAAI;AACJ,IAAI,IAAI,GAAG;AACX,QAAQ,MAAM,IAAI,CAAC,aAAa,CAAC,iDAAiD,CAAC;AACnF,IAAI;AACJ,IAAI,qBAAqB,GAAG;AAC5B,QAAQ,MAAM,IAAI,CAAC,aAAa,CAAC,iDAAiD,CAAC;AACnF,IAAI;AACJ,IAAI,WAAW,GAAG;AAClB,QAAQ,MAAM,IAAI,CAAC,aAAa,CAAC,iDAAiD,CAAC;AACnF,IAAI;AACJ,IAAI,gBAAgB,GAAG;AACvB,QAAQ,MAAM,IAAI,CAAC,aAAa,CAAC,4DAA4D,CAAC;AAC9F,IAAI;AACJ,IAAI,kBAAkB,GAAG;AACzB,QAAQ,MAAM,IAAI,CAAC,aAAa,CAAC,4DAA4D,CAAC;AAC9F,IAAI;AACJ;;;;;;;;;"}

package/dist/plugin.js

@@ -0,0 +1,42 @@
+var capacitorCapacitorSpeechRecognition = (function (exports, core) {
+    'use strict';
+
+    const SpeechRecognition = core.registerPlugin('SpeechRecognition', {
+        web: () => Promise.resolve().then(function () { return web; }).then((m) => new m.SpeechRecognitionWeb()),
+    });
+
+    class SpeechRecognitionWeb extends core.WebPlugin {
+        available() {
+            throw this.unimplemented('Speech recognition is not available on the web.');
+        }
+        start(_options) {
+            throw this.unimplemented('Speech recognition is not available on the web.');
+        }
+        stop() {
+            throw this.unimplemented('Speech recognition is not available on the web.');
+        }
+        getSupportedLanguages() {
+            throw this.unimplemented('Speech recognition is not available on the web.');
+        }
+        isListening() {
+            throw this.unimplemented('Speech recognition is not available on the web.');
+        }
+        checkPermissions() {
+            throw this.unimplemented('Speech recognition permissions are not handled on the web.');
+        }
+        requestPermissions() {
+            throw this.unimplemented('Speech recognition permissions are not handled on the web.');
+        }
+    }
+
+    var web = /*#__PURE__*/Object.freeze({
+        __proto__: null,
+        SpeechRecognitionWeb: SpeechRecognitionWeb
+    });
+
+    exports.SpeechRecognition = SpeechRecognition;
+
+    return exports;
+
+})({}, capacitorExports);
+//# sourceMappingURL=plugin.js.map

package/dist/plugin.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"plugin.js","sources":["esm/index.js","esm/web.js"],"sourcesContent":["import { registerPlugin } from '@capacitor/core';\nconst SpeechRecognition = registerPlugin('SpeechRecognition', {\n    web: () => import('./web').then((m) => new m.SpeechRecognitionWeb()),\n});\nexport * from './definitions';\nexport { SpeechRecognition };\n//# sourceMappingURL=index.js.map","import { WebPlugin } from '@capacitor/core';\nexport class SpeechRecognitionWeb extends WebPlugin {\n    available() {\n        throw this.unimplemented('Speech recognition is not available on the web.');\n    }\n    start(_options) {\n        throw this.unimplemented('Speech recognition is not available on the web.');\n    }\n    stop() {\n        throw this.unimplemented('Speech recognition is not available on the web.');\n    }\n    getSupportedLanguages() {\n        throw this.unimplemented('Speech recognition is not available on the web.');\n    }\n    isListening() {\n        throw this.unimplemented('Speech recognition is not available on the web.');\n    }\n    checkPermissions() {\n        throw this.unimplemented('Speech recognition permissions are not handled on the web.');\n    }\n    requestPermissions() {\n        throw this.unimplemented('Speech recognition permissions are not handled on the web.');\n    }\n}\n//# sourceMappingURL=web.js.map"],"names":["registerPlugin","WebPlugin"],"mappings":";;;AACK,UAAC,iBAAiB,GAAGA,mBAAc,CAAC,mBAAmB,EAAE;IAC9D,IAAI,GAAG,EAAE,MAAM,mDAAe,CAAC,IAAI,CAAC,CAAC,CAAC,KAAK,IAAI,CAAC,CAAC,oBAAoB,EAAE,CAAC;IACxE,CAAC;;ICFM,MAAM,oBAAoB,SAASC,cAAS,CAAC;IACpD,IAAI,SAAS,GAAG;IAChB,QAAQ,MAAM,IAAI,CAAC,aAAa,CAAC,iDAAiD,CAAC;IACnF,IAAI;IACJ,IAAI,KAAK,CAAC,QAAQ,EAAE;IACpB,QAAQ,MAAM,IAAI,CAAC,aAAa,CAAC,iDAAiD,CAAC;IACnF,IAAI;IACJ,IAAI,IAAI,GAAG;IACX,QAAQ,MAAM,IAAI,CAAC,aAAa,CAAC,iDAAiD,CAAC;IACnF,IAAI;IACJ,IAAI,qBAAqB,GAAG;IAC5B,QAAQ,MAAM,IAAI,CAAC,aAAa,CAAC,iDAAiD,CAAC;IACnF,IAAI;IACJ,IAAI,WAAW,GAAG;IAClB,QAAQ,MAAM,IAAI,CAAC,aAAa,CAAC,iDAAiD,CAAC;IACnF,IAAI;IACJ,IAAI,gBAAgB,GAAG;IACvB,QAAQ,MAAM,IAAI,CAAC,aAAa,CAAC,4DAA4D,CAAC;IAC9F,IAAI;IACJ,IAAI,kBAAkB,GAAG;IACzB,QAAQ,MAAM,IAAI,CAAC,aAAa,CAAC,4DAA4D,CAAC;IAC9F,IAAI;IACJ;;;;;;;;;;;;;;;"}

package/ios/Sources/SpeechRecognitionPlugin/SpeechRecognitionPlugin.swift

@@ -0,0 +1,282 @@
+import AVFoundation
+import Capacitor
+import Foundation
+import Speech
+
+private enum PermissionState: String {
+    case granted
+    case denied
+    case prompt
+}
+
+@objc(SpeechRecognitionPlugin)
+public final class SpeechRecognitionPlugin: CAPPlugin, CAPBridgedPlugin {
+    public let identifier = "SpeechRecognitionPlugin"
+    public let jsName = "SpeechRecognition"
+    public let pluginMethods: [CAPPluginMethod] = [
+        CAPPluginMethod(name: "available", returnType: CAPPluginReturnPromise),
+        CAPPluginMethod(name: "start", returnType: CAPPluginReturnPromise),
+        CAPPluginMethod(name: "stop", returnType: CAPPluginReturnPromise),
+        CAPPluginMethod(name: "getSupportedLanguages", returnType: CAPPluginReturnPromise),
+        CAPPluginMethod(name: "isListening", returnType: CAPPluginReturnPromise),
+        CAPPluginMethod(name: "checkPermissions", returnType: CAPPluginReturnPromise),
+        CAPPluginMethod(name: "requestPermissions", returnType: CAPPluginReturnPromise)
+    ]
+
+    private let audioEngine = AVAudioEngine()
+    private var recognitionRequest: SFSpeechAudioBufferRecognitionRequest?
+    private var recognitionTask: SFSpeechRecognitionTask?
+    private var speechRecognizer: SFSpeechRecognizer?
+    private var activeCall: CAPPluginCall?
+    private var currentOptions: RecognitionOptions?
+    private var hasInstalledTap = false
+
+    private let maxDefaultResults = 5
+
+    @objc func available(_ call: CAPPluginCall) {
+        let locale = Locale(identifier: call.getString("language") ?? Locale.current.identifier)
+        let recognizer = SFSpeechRecognizer(locale: locale)
+        call.resolve(["available": recognizer?.isAvailable ?? false])
+    }
+
+    @objc func start(_ call: CAPPluginCall) {
+        if self.audioEngine.isRunning || recognitionTask != nil {
+            call.reject("Speech recognition is already running.")
+            return
+        }
+
+        guard isSpeechPermissionGranted else {
+            call.reject("Missing speech recognition permission.")
+            return
+        }
+
+        let options = RecognitionOptions(
+            language: call.getString("language") ?? Locale.current.identifier,
+            maxResults: call.getInt("maxResults") ?? maxDefaultResults,
+            partialResults: call.getBool("partialResults") ?? false,
+            addPunctuation: call.getBool("addPunctuation") ?? false
+        )
+
+        self.activeCall = call
+        self.currentOptions = options
+
+        AVAudioSession.sharedInstance().requestRecordPermission { granted in
+            guard granted else {
+                DispatchQueue.main.async {
+                    call.reject("User denied microphone access.")
+                    self.cleanupRecognition(notifyStop: false)
+                }
+                return
+            }
+
+            DispatchQueue.main.async {
+                self.beginRecognition(call: call, options: options)
+            }
+        }
+    }
+
+    @objc func stop(_ call: CAPPluginCall) {
+        cleanupRecognition(notifyStop: true)
+        call.resolve()
+    }
+
+    @objc func isListening(_ call: CAPPluginCall) {
+        call.resolve(["listening": audioEngine.isRunning])
+    }
+
+    @objc func getSupportedLanguages(_ call: CAPPluginCall) {
+        let identifiers = SFSpeechRecognizer
+            .supportedLocales()
+            .map { $0.identifier }
+            .sorted()
+        call.resolve(["languages": identifiers])
+    }
+
+    @objc override public func checkPermissions(_ call: CAPPluginCall) {
+        call.resolve(["speechRecognition": permissionState.rawValue])
+    }
+
+    @objc override public func requestPermissions(_ call: CAPPluginCall) {
+        SFSpeechRecognizer.requestAuthorization { status in
+            switch status {
+            case .authorized:
+                AVAudioSession.sharedInstance().requestRecordPermission { granted in
+                    DispatchQueue.main.async {
+                        let result: PermissionState = granted ? .granted : .denied
+                        call.resolve(["speechRecognition": result.rawValue])
+                    }
+                }
+            case .denied, .restricted:
+                DispatchQueue.main.async {
+                    call.resolve(["speechRecognition": PermissionState.denied.rawValue])
+                }
+            case .notDetermined:
+                DispatchQueue.main.async {
+                    call.resolve(["speechRecognition": PermissionState.prompt.rawValue])
+                }
+            @unknown default:
+                DispatchQueue.main.async {
+                    call.resolve(["speechRecognition": PermissionState.prompt.rawValue])
+                }
+            }
+        }
+    }
+
+    private func beginRecognition(call: CAPPluginCall, options: RecognitionOptions) {
+        guard let recognizer = SFSpeechRecognizer(locale: Locale(identifier: options.language)) else {
+            call.reject("Unsupported locale: \(options.language)")
+            cleanupRecognition(notifyStop: false)
+            return
+        }
+
+        guard recognizer.isAvailable else {
+            call.reject("Speech recognizer is currently unavailable.")
+            cleanupRecognition(notifyStop: false)
+            return
+        }
+
+        speechRecognizer = recognizer
+
+        do {
+            try configureAudioSession()
+        } catch {
+            call.reject("Failed to configure audio session: \(error.localizedDescription)")
+            cleanupRecognition(notifyStop: false)
+            return
+        }
+
+        let recognitionRequest = SFSpeechAudioBufferRecognitionRequest()
+        recognitionRequest.shouldReportPartialResults = options.partialResults
+        if #available(iOS 16.0, *) {
+            recognitionRequest.addsPunctuation = options.addPunctuation
+        }
+        self.recognitionRequest = recognitionRequest
+
+        let inputNode = audioEngine.inputNode
+        let recordingFormat = inputNode.outputFormat(forBus: 0)
+        inputNode.removeTap(onBus: 0)
+        inputNode.installTap(onBus: 0, bufferSize: 1024, format: recordingFormat) { [weak self] buffer, _ in
+            self?.recognitionRequest?.append(buffer)
+        }
+        hasInstalledTap = true
+
+        audioEngine.prepare()
+
+        do {
+            try audioEngine.start()
+            notifyListeners("listeningState", data: ["status": "started"])
+        } catch {
+            call.reject("Unable to start audio engine: \(error.localizedDescription)")
+            cleanupRecognition(notifyStop: false)
+            return
+        }
+
+        if options.partialResults {
+            call.resolve()
+        }
+
+        recognitionTask = recognizer.recognitionTask(with: recognitionRequest) { [weak self] result, error in
+            guard let self else { return }
+            if let result {
+                let matches = self.buildMatches(from: result, maxResults: options.maxResults)
+                if options.partialResults {
+                    DispatchQueue.main.async {
+                        self.notifyListeners("partialResults", data: ["matches": matches])
+                    }
+                } else if result.isFinal {
+                    DispatchQueue.main.async {
+                        self.activeCall?.resolve(["matches": matches])
+                    }
+                }
+
+                if result.isFinal {
+                    self.cleanupRecognition(notifyStop: true)
+                }
+            }
+
+            if let error {
+                self.handleRecognitionError(error)
+            }
+        }
+    }
+
+    private func configureAudioSession() throws {
+        let session = AVAudioSession.sharedInstance()
+        try session.setCategory(.playAndRecord, options: [.defaultToSpeaker, .duckOthers])
+        try session.setMode(.measurement)
+        try session.setActive(true, options: .notifyOthersOnDeactivation)
+    }
+
+    private func cleanupRecognition(notifyStop: Bool) {
+        DispatchQueue.main.async {
+            if self.audioEngine.isRunning {
+                self.audioEngine.stop()
+            }
+
+            if self.hasInstalledTap {
+                self.audioEngine.inputNode.removeTap(onBus: 0)
+                self.hasInstalledTap = false
+            }
+
+            self.recognitionRequest?.endAudio()
+            self.recognitionRequest = nil
+            self.recognitionTask?.cancel()
+            self.recognitionTask = nil
+            self.speechRecognizer = nil
+            self.currentOptions = nil
+            self.activeCall = nil
+
+            if notifyStop {
+                self.notifyListeners("listeningState", data: ["status": "stopped"])
+            }
+        }
+    }
+
+    private func handleRecognitionError(_ error: Error) {
+        DispatchQueue.main.async {
+            self.cleanupRecognition(notifyStop: true)
+            self.activeCall?.reject(error.localizedDescription)
+        }
+    }
+
+    private func buildMatches(from result: SFSpeechRecognitionResult, maxResults: Int) -> [String] {
+        var matches: [String] = []
+        for transcription in result.transcriptions where matches.count < maxResults {
+            matches.append(transcription.formattedString)
+        }
+        return matches
+    }
+
+    private var isSpeechPermissionGranted: Bool {
+        switch SFSpeechRecognizer.authorizationStatus() {
+        case .authorized:
+            return true
+        case .notDetermined, .denied, .restricted:
+            return false
+        @unknown default:
+            return false
+        }
+    }
+
+    private var permissionState: PermissionState {
+        let speechStatus = SFSpeechRecognizer.authorizationStatus()
+        let micStatus = AVAudioSession.sharedInstance().recordPermission
+
+        if speechStatus == .denied || speechStatus == .restricted || micStatus == .denied {
+            return .denied
+        }
+
+        if speechStatus == .notDetermined || micStatus == .undetermined {
+            return .prompt
+        }
+
+        return .granted
+    }
+}
+
+private struct RecognitionOptions {
+    let language: String
+    let maxResults: Int
+    let partialResults: Bool
+    let addPunctuation: Bool
+}

package/ios/Tests/SpeechRecognitionPluginTests/SpeechRecognitionPluginTests.swift

@@ -0,0 +1,7 @@
+import XCTest
+
+final class SpeechRecognitionPluginTests: XCTestCase {
+    func testExample() {
+        XCTAssertTrue(true)
+    }
+}

package/package.json

@@ -0,0 +1,87 @@
+{
+  "name": "@capgo/capacitor-speech-recognition",
+  "version": "7.0.0",
+  "description": "Modern Capacitor plugin for on-device and cloud speech recognition with live partial results.",
+  "main": "dist/plugin.cjs.js",
+  "module": "dist/esm/index.js",
+  "types": "dist/esm/index.d.ts",
+  "unpkg": "dist/plugin.js",
+  "files": [
+    "android/src/main/",
+    "android/build.gradle",
+    "dist/",
+    "ios/Sources",
+    "ios/Tests",
+    "Package.swift",
+    "CapgoCapacitorSpeechRecognition.podspec"
+  ],
+  "author": "Martin Donadieu <martin@capgo.app>",
+  "license": "MIT",
+  "repository": {
+    "type": "git",
+    "url": "git+https://github.com/Cap-go/capacitor-speech-recognition.git"
+  },
+  "bugs": {
+    "url": "https://github.com/Cap-go/capacitor-speech-recognition/issues"
+  },
+  "keywords": [
+    "capacitor",
+    "plugin",
+    "speech",
+    "recognition",
+    "voice",
+    "transcription"
+  ],
+  "scripts": {
+    "verify": "npm run verify:ios && npm run verify:android && npm run verify:web",
+    "verify:ios": "xcodebuild -scheme CapgoCapacitorSpeechRecognition -destination generic/platform=iOS",
+    "verify:android": "cd android && ./gradlew clean build test && cd ..",
+    "verify:web": "npm run build",
+    "lint": "npm run eslint && npm run prettier -- --check && npm run swiftlint -- lint",
+    "fmt": "npm run eslint -- --fix && npm run prettier -- --write && npm run swiftlint -- --fix --format",
+    "eslint": "eslint . --ext .ts",
+    "prettier": "prettier \"**/*.{css,html,ts,js,java,kt}\" --plugin=prettier-plugin-java",
+    "swiftlint": "node-swiftlint",
+    "docgen": "docgen --api SpeechRecognitionPlugin --output-readme README.md --output-json dist/docs.json",
+    "build": "npm run clean && npm run docgen && tsc && rollup -c rollup.config.mjs",
+    "clean": "rimraf ./dist",
+    "watch": "tsc --watch",
+    "prepublishOnly": "npm run build"
+  },
+  "devDependencies": {
+    "@capacitor/android": "^7.0.0",
+    "@capacitor/cli": "^7.0.0",
+    "@capacitor/core": "^7.0.0",
+    "@capacitor/docgen": "^0.3.0",
+    "@capacitor/ios": "^7.0.0",
+    "@ionic/eslint-config": "^0.4.0",
+    "@ionic/prettier-config": "^4.0.0",
+    "@ionic/swiftlint-config": "^2.0.0",
+    "@types/node": "^22.13.1",
+    "eslint": "^8.57.0",
+    "eslint-plugin-import": "^2.31.0",
+    "husky": "^9.1.7",
+    "prettier": "^3.4.2",
+    "prettier-plugin-java": "^2.6.7",
+    "rimraf": "^6.0.1",
+    "rollup": "^4.34.6",
+    "swiftlint": "^2.0.0",
+    "typescript": "^5.7.3"
+  },
+  "peerDependencies": {
+    "@capacitor/core": ">=7.0.0"
+  },
+  "eslintConfig": {
+    "extends": "@ionic/eslint-config/recommended"
+  },
+  "prettier": "@ionic/prettier-config",
+  "swiftlint": "@ionic/swiftlint-config",
+  "capacitor": {
+    "ios": {
+      "src": "ios"
+    },
+    "android": {
+      "src": "android"
+    }
+  }
+}