@myscheme/voice-navigation-sdk 0.1.8 → 0.1.10

package/dist/server/security.js

@@ -0,0 +1,89 @@
+const rateLimitMap = new Map();
+const MAX_TRACKED_CLIENTS = 20000;
+function headerString(value) {
+    if (Array.isArray(value)) {
+        return value[0] ?? "";
+    }
+    return value ?? "";
+}
+function hostOf(urlOrOrigin) {
+    try {
+        return new URL(urlOrOrigin).host;
+    }
+    catch {
+        return null;
+    }
+}
+export function isTrustProxyEnabled() {
+    return String(process.env.VOICE_NAV_TRUST_PROXY || "").toLowerCase() === "true";
+}
+export function parseAllowedOrigins() {
+    return String(process.env.VOICE_NAV_ALLOWED_ORIGINS || "")
+        .split(",")
+        .map((entry) => entry.trim())
+        .filter(Boolean);
+}
+export function getClientId(req) {
+    if (isTrustProxyEnabled()) {
+        const forwardedFor = headerString(req.headers?.["x-forwarded-for"]);
+        if (forwardedFor) {
+            return forwardedFor.split(",")[0].trim();
+        }
+        const realIp = headerString(req.headers?.["x-real-ip"]);
+        if (realIp) {
+            return realIp;
+        }
+    }
+    return req.socket?.remoteAddress || "unknown";
+}
+export function isOriginAllowed(req) {
+    const origin = headerString(req.headers?.origin);
+    const referer = headerString(req.headers?.referer);
+    const host = headerString(req.headers?.host);
+    const candidateHost = origin
+        ? hostOf(origin)
+        : referer
+            ? hostOf(referer)
+            : null;
+    if (!candidateHost) {
+        return false;
+    }
+    const allowed = parseAllowedOrigins();
+    if (allowed.length > 0) {
+        return allowed.some((entry) => (hostOf(entry) || entry) === candidateHost);
+    }
+    return host ? candidateHost === host : false;
+}
+export function checkRateLimit(identifier, maxRequests, windowMs) {
+    const now = Date.now();
+    if (rateLimitMap.size > MAX_TRACKED_CLIENTS) {
+        for (const [key, record] of rateLimitMap) {
+            if (now > record.resetTime) {
+                rateLimitMap.delete(key);
+            }
+        }
+        if (rateLimitMap.size > MAX_TRACKED_CLIENTS) {
+            rateLimitMap.clear();
+        }
+    }
+    const record = rateLimitMap.get(identifier);
+    if (!record || now > record.resetTime) {
+        rateLimitMap.set(identifier, { count: 1, resetTime: now + windowMs });
+        return true;
+    }
+    if (record.count >= maxRequests) {
+        return false;
+    }
+    record.count++;
+    return true;
+}
+export function evaluateGate(req, options) {
+    if (!isOriginAllowed(req)) {
+        return { ok: false, status: 403, error: "Origin not allowed" };
+    }
+    const key = `${options.bucket ?? "default"}:${getClientId(req)}`;
+    if (!checkRateLimit(key, options.maxRequests, options.windowMs)) {
+        return { ok: false, status: 429, error: "Too many requests" };
+    }
+    return { ok: true };
+}

package/dist/services/bedrock.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"bedrock.d.ts","sourceRoot":"","sources":["../../src/services/bedrock.ts"],"names":[],"mappings":"AASA,OAAO,KAAK,EAAE,mBAAmB,EAAmB,MAAM,aAAa,CAAC;AACxE,OAAO,KAAK,EAAE,YAAY,EAAE,MAAM,oBAAoB,CAAC;AAEvD,MAAM,WAAW,aAAa;IAE5B,eAAe,CAAC,EAAE,MAAM,CAAC;IACzB,iBAAiB,CAAC,EAAE,MAAM,CAAC;IAG3B,MAAM,CAAC,EAAE,MAAM,CAAC;IAChB,WAAW,CAAC,EAAE,MAAM,CAAC;IACrB,eAAe,CAAC,EAAE,MAAM,CAAC;IACzB,OAAO,CAAC,EAAE,MAAM,CAAC;IACjB,gBAAgB,CAAC,EAAE,MAAM,GAAG,IAAI,CAAC;CAClC;AAqDD,qBAAa,cAAc;IACzB,OAAO,CAAC,MAAM,CAAqC;IACnD,OAAO,CAAC,aAAa,CAAuB;IAC5C,OAAO,CAAC,gBAAgB,CAAgB;IACxC,OAAO,CAAC,YAAY,CAA6B;IACjD,OAAO,CAAC,eAAe,CAAuB;IAC9C,OAAO,CAAC,oBAAoB,CAAuB;IACnD,OAAO,CAAC,QAAQ,CAAkB;gBAEtB,MAAM,EAAE,aAAa;IA6CjC,eAAe,CAAC,QAAQ,EAAE,YAAY,GAAG,IAAI;IAO7C,OAAO,CAAC,iBAAiB;IA+BnB,aAAa,CAAC,SAAS,EAAE,MAAM,GAAG,OAAO,CAAC,mBAAmB,CAAC;YAWtD,qBAAqB;YA2ErB,mBAAmB;IAyEjC,MAAM,CAAC,MAAM,CAAC,MAAM,EAAE,aAAa,GAAG,cAAc;IAO9C,SAAS,CAAC,IAAI,EAAE,MAAM,GAAG,OAAO,CAAC,MAAM,EAAE,CAAC;YAWlC,iBAAiB;YAkFjB,eAAe;IAwFvB,uBAAuB,CAC3B,MAAM,EAAE,MAAM,EACd,QAAQ,EAAE,MAAM,GACf,OAAO,CAAC,MAAM,CAAC;YA0BJ,wBAAwB;YAwExB,sBAAsB;CA+DrC"}
+{"version":3,"file":"bedrock.d.ts","sourceRoot":"","sources":["../../src/services/bedrock.ts"],"names":[],"mappings":"AASA,OAAO,KAAK,EAAE,mBAAmB,EAAmB,MAAM,aAAa,CAAC;AACxE,OAAO,KAAK,EAAE,YAAY,EAAE,MAAM,oBAAoB,CAAC;AAEvD,MAAM,WAAW,aAAa;IAE5B,eAAe,CAAC,EAAE,MAAM,CAAC;IACzB,iBAAiB,CAAC,EAAE,MAAM,CAAC;IAG3B,MAAM,CAAC,EAAE,MAAM,CAAC;IAChB,WAAW,CAAC,EAAE,MAAM,CAAC;IACrB,eAAe,CAAC,EAAE,MAAM,CAAC;IACzB,OAAO,CAAC,EAAE,MAAM,CAAC;IACjB,gBAAgB,CAAC,EAAE,MAAM,GAAG,IAAI,CAAC;CAClC;AAiED,qBAAa,cAAc;IACzB,OAAO,CAAC,MAAM,CAAqC;IACnD,OAAO,CAAC,aAAa,CAAuB;IAC5C,OAAO,CAAC,gBAAgB,CAAgB;IACxC,OAAO,CAAC,YAAY,CAA6B;IACjD,OAAO,CAAC,eAAe,CAAuB;IAC9C,OAAO,CAAC,oBAAoB,CAAuB;IACnD,OAAO,CAAC,QAAQ,CAAkB;gBAEtB,MAAM,EAAE,aAAa;IA6CjC,eAAe,CAAC,QAAQ,EAAE,YAAY,GAAG,IAAI;IAO7C,OAAO,CAAC,iBAAiB;IA+BnB,aAAa,CAAC,SAAS,EAAE,MAAM,GAAG,OAAO,CAAC,mBAAmB,CAAC;YAWtD,qBAAqB;YA2ErB,mBAAmB;IAyEjC,MAAM,CAAC,MAAM,CAAC,MAAM,EAAE,aAAa,GAAG,cAAc;IAO9C,SAAS,CAAC,IAAI,EAAE,MAAM,GAAG,OAAO,CAAC,MAAM,EAAE,CAAC;YAWlC,iBAAiB;YAkFjB,eAAe;IAwFvB,uBAAuB,CAC3B,MAAM,EAAE,MAAM,EACd,QAAQ,EAAE,MAAM,GACf,OAAO,CAAC,MAAM,CAAC;YA2BJ,wBAAwB;YAwExB,sBAAsB;CA+DrC"}

package/dist/services/bedrock.js

@@ -11,12 +11,24 @@ CRITICAL RULES:
 6. FOR PAGE NAVIGATION, USE THE EXACT "navigate_<id>" ACTION PROVIDED FOR THAT PAGE
 7. IF UNCERTAIN, RESPOND WITH {"action": "unknown"}
 
+MULTILINGUAL INPUT:
+- The user may speak in ANY supported Indian language (e.g. Hindi, Tamil, Telugu, Bengali, Marathi, Gujarati, Kannada, Malayalam, Punjabi, Odia, Assamese, Urdu, and others), or mix languages.
+- Understand the MEANING of the command regardless of the input language and map it to the correct action.
+- The "action" value, action IDs, and "navigate_<id>" page IDs are ALWAYS in English/ASCII and MUST NOT be translated or transliterated.
+- For "query" (search) and "text" (type) fields, preserve the user's own words as spoken.
+
 RESPONSE FORMAT:
 - Core actions: {"action": "action_name"}
 - Search: {"action": "search_content", "query": "keywords"}
 - Type text: {"action": "type_text", "text": "text to type"}
 - Page navigation: {"action": "navigate_<page_id>"}
 
+MULTILINGUAL EXAMPLES:
+- User says "नीचे स्क्रॉल करो" (Hindi: scroll down) → {"action": "scroll_down"}
+- User says "डैशबोर्ड खोलो" (Hindi: open dashboard) → {"action": "navigate_dashboard"}
+- User says "மேலே செல்" (Tamil: go up) → {"action": "scroll_up"}
+- User says "వెనక్కి వెళ్ళు" (Telugu: go back) → {"action": "go_back"}
+
 EXAMPLES:
 - User says "go to dashboard" → {"action": "navigate_dashboard"}
 - User says "show me the team page" → {"action": "navigate_team"}
@@ -362,6 +374,7 @@ export class BedrockService {
 Generate brief, natural, conversational feedback messages (maximum 10 words).
 Be specific and clear about what action was performed.
 Use the language: ${language}
+Never output internal action identifiers, underscores, technical codes, or JSON — describe the action only in plain, everyday words a non-technical listener would understand.
 Respond with ONLY the feedback message text, no JSON, no explanations.`;
         const userPrompt = prompt;
         console.log("[Bedrock] Using proxy mode:", this.useProxy);

package/dist/services/sarvam-speech.d.ts

@@ -0,0 +1,47 @@
+export interface SarvamSpeechConfig {
+    ttsUrl?: string;
+    sttUrl?: string;
+    sttModel?: "saaras:v3" | "saarika:v2.5";
+    ttsModel?: "bulbul:v3" | "bulbul:v2";
+    speaker?: string;
+    sttSampleRate?: 8000 | 16000;
+    ttsSampleRate?: 8000 | 16000 | 22050 | 24000 | 32000 | 44100 | 48000;
+}
+export interface SarvamSttHandlers {
+    onPartial?: (text: string) => void;
+    onFinal?: (text: string) => void;
+    onSpeechStart?: () => void;
+    onSpeechEnd?: () => void;
+    onError?: (error: Error) => void;
+    shouldSendAudio?: () => boolean;
+}
+export declare class SarvamSpeechService {
+    private readonly ttsUrl;
+    private readonly sttUrl;
+    private readonly sttModel;
+    private readonly ttsModel;
+    private readonly speaker;
+    private readonly sttSampleRate;
+    private readonly ttsSampleRate;
+    private sttSocket;
+    private sttContext;
+    private sttMediaStream;
+    private sttSourceNode;
+    private sttProcessorNode;
+    private audioElement;
+    private currentBlobUrl;
+    private speechGeneration;
+    private listeningGeneration;
+    constructor(config?: SarvamSpeechConfig);
+    startListening(language: string, handlers: SarvamSttHandlers): Promise<void>;
+    private sendFlush;
+    stopListening(): Promise<void>;
+    speak(text: string, opts?: {
+        language?: string;
+        onAudioStart?: () => void;
+    }): Promise<void>;
+    stopSpeaking(): void;
+    dispose(): Promise<void>;
+    private cleanupAudioElement;
+}
+//# sourceMappingURL=sarvam-speech.d.ts.map

package/dist/services/sarvam-speech.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"sarvam-speech.d.ts","sourceRoot":"","sources":["../../src/services/sarvam-speech.ts"],"names":[],"mappings":"AAAA,MAAM,WAAW,kBAAkB;IACjC,MAAM,CAAC,EAAE,MAAM,CAAC;IAChB,MAAM,CAAC,EAAE,MAAM,CAAC;IAChB,QAAQ,CAAC,EAAE,WAAW,GAAG,cAAc,CAAC;IACxC,QAAQ,CAAC,EAAE,WAAW,GAAG,WAAW,CAAC;IACrC,OAAO,CAAC,EAAE,MAAM,CAAC;IACjB,aAAa,CAAC,EAAE,IAAI,GAAG,KAAK,CAAC;IAC7B,aAAa,CAAC,EAAE,IAAI,GAAG,KAAK,GAAG,KAAK,GAAG,KAAK,GAAG,KAAK,GAAG,KAAK,GAAG,KAAK,CAAC;CACtE;AAED,MAAM,WAAW,iBAAiB;IAChC,SAAS,CAAC,EAAE,CAAC,IAAI,EAAE,MAAM,KAAK,IAAI,CAAC;IACnC,OAAO,CAAC,EAAE,CAAC,IAAI,EAAE,MAAM,KAAK,IAAI,CAAC;IACjC,aAAa,CAAC,EAAE,MAAM,IAAI,CAAC;IAC3B,WAAW,CAAC,EAAE,MAAM,IAAI,CAAC;IACzB,OAAO,CAAC,EAAE,CAAC,KAAK,EAAE,KAAK,KAAK,IAAI,CAAC;IACjC,eAAe,CAAC,EAAE,MAAM,OAAO,CAAC;CACjC;AAmHD,qBAAa,mBAAmB;IAC9B,OAAO,CAAC,QAAQ,CAAC,MAAM,CAAS;IAChC,OAAO,CAAC,QAAQ,CAAC,MAAM,CAAS;IAChC,OAAO,CAAC,QAAQ,CAAC,QAAQ,CAA+B;IACxD,OAAO,CAAC,QAAQ,CAAC,QAAQ,CAA4B;IACrD,OAAO,CAAC,QAAQ,CAAC,OAAO,CAAS;IACjC,OAAO,CAAC,QAAQ,CAAC,aAAa,CAAe;IAC7C,OAAO,CAAC,QAAQ,CAAC,aAAa,CAQhB;IAEd,OAAO,CAAC,SAAS,CAA0B;IAC3C,OAAO,CAAC,UAAU,CAA6B;IAC/C,OAAO,CAAC,cAAc,CAA4B;IAClD,OAAO,CAAC,aAAa,CAA2C;IAChE,OAAO,CAAC,gBAAgB,CAAoC;IAE5D,OAAO,CAAC,YAAY,CAAiC;IACrD,OAAO,CAAC,cAAc,CAAuB;IAC7C,OAAO,CAAC,gBAAgB,CAAa;IACrC,OAAO,CAAC,mBAAmB,CAAa;gBAE5B,MAAM,GAAE,kBAAuB;IAUrC,cAAc,CAClB,QAAQ,EAAE,MAAM,EAChB,QAAQ,EAAE,iBAAiB,GAC1B,OAAO,CAAC,IAAI,CAAC;IA+KhB,OAAO,CAAC,SAAS;IAWX,aAAa,IAAI,OAAO,CAAC,IAAI,CAAC;IAiE9B,KAAK,CACT,IAAI,EAAE,MAAM,EACZ,IAAI,CAAC,EAAE;QAAE,QAAQ,CAAC,EAAE,MAAM,CAAC;QAAC,YAAY,CAAC,EAAE,MAAM,IAAI,CAAA;KAAE,GACtD,OAAO,CAAC,IAAI,CAAC;IAqGhB,YAAY,IAAI,IAAI;IAuBd,OAAO,IAAI,OAAO,CAAC,IAAI,CAAC;IAK9B,OAAO,CAAC,mBAAmB;CAc5B"}

package/dist/services/sarvam-speech.js

@@ -0,0 +1,418 @@
+const DEFAULT_TTS_URL = "/api/speech/sarvam";
+const DEFAULT_STT_URL = "/api/speech/sarvam-stt";
+const DEFAULT_STT_MODEL = "saaras:v3";
+const DEFAULT_TTS_MODEL = "bulbul:v3";
+const DEFAULT_SPEAKER = "manan";
+const DEFAULT_STT_SAMPLE_RATE = 16000;
+function toBase64(bytes) {
+    let binary = "";
+    const chunkSize = 0x8000;
+    for (let i = 0; i < bytes.length; i += chunkSize) {
+        binary += String.fromCharCode(...bytes.subarray(i, i + chunkSize));
+    }
+    return btoa(binary);
+}
+function downsampleTo16BitMono(input, inputRate, outputRate) {
+    if (outputRate === inputRate) {
+        const pcm = new Int16Array(input.length);
+        for (let i = 0; i < input.length; i++) {
+            const clamped = Math.max(-1, Math.min(1, input[i]));
+            pcm[i] = clamped < 0 ? clamped * 0x8000 : clamped * 0x7fff;
+        }
+        return pcm;
+    }
+    const ratio = inputRate / outputRate;
+    const outputLength = Math.max(1, Math.floor(input.length / ratio));
+    const pcm = new Int16Array(outputLength);
+    let outputIndex = 0;
+    let inputIndex = 0;
+    while (outputIndex < outputLength) {
+        const nextInputIndex = Math.floor((outputIndex + 1) * ratio);
+        let accum = 0;
+        let count = 0;
+        for (let i = inputIndex; i < nextInputIndex && i < input.length; i++) {
+            accum += input[i];
+            count++;
+        }
+        const sample = count > 0 ? accum / count : 0;
+        const clamped = Math.max(-1, Math.min(1, sample));
+        pcm[outputIndex] = clamped < 0 ? clamped * 0x8000 : clamped * 0x7fff;
+        outputIndex++;
+        inputIndex = nextInputIndex;
+    }
+    return pcm;
+}
+function encodeWav(pcm, sampleRate) {
+    const bytesPerSample = 2;
+    const numChannels = 1;
+    const dataSize = pcm.length * bytesPerSample;
+    const buffer = new ArrayBuffer(44 + dataSize);
+    const view = new DataView(buffer);
+    const writeString = (offset, value) => {
+        for (let i = 0; i < value.length; i++) {
+            view.setUint8(offset + i, value.charCodeAt(i));
+        }
+    };
+    writeString(0, "RIFF");
+    view.setUint32(4, 36 + dataSize, true);
+    writeString(8, "WAVE");
+    writeString(12, "fmt ");
+    view.setUint32(16, 16, true);
+    view.setUint16(20, 1, true);
+    view.setUint16(22, numChannels, true);
+    view.setUint32(24, sampleRate, true);
+    view.setUint32(28, sampleRate * numChannels * bytesPerSample, true);
+    view.setUint16(32, numChannels * bytesPerSample, true);
+    view.setUint16(34, 16, true);
+    writeString(36, "data");
+    view.setUint32(40, dataSize, true);
+    let offset = 44;
+    for (let i = 0; i < pcm.length; i++) {
+        view.setInt16(offset, pcm[i], true);
+        offset += 2;
+    }
+    return new Uint8Array(buffer);
+}
+function resolveWsUrl(path) {
+    if (/^wss?:\/\//i.test(path)) {
+        return path;
+    }
+    if (typeof window === "undefined") {
+        return path;
+    }
+    const protocol = window.location.protocol === "https:" ? "wss:" : "ws:";
+    return `${protocol}//${window.location.host}${path}`;
+}
+export class SarvamSpeechService {
+    constructor(config = {}) {
+        this.sttSocket = null;
+        this.sttContext = null;
+        this.sttMediaStream = null;
+        this.sttSourceNode = null;
+        this.sttProcessorNode = null;
+        this.audioElement = null;
+        this.currentBlobUrl = null;
+        this.speechGeneration = 0;
+        this.listeningGeneration = 0;
+        this.ttsUrl = config.ttsUrl ?? DEFAULT_TTS_URL;
+        this.sttUrl = config.sttUrl ?? DEFAULT_STT_URL;
+        this.sttModel = config.sttModel ?? DEFAULT_STT_MODEL;
+        this.ttsModel = config.ttsModel ?? DEFAULT_TTS_MODEL;
+        this.speaker = config.speaker ?? DEFAULT_SPEAKER;
+        this.sttSampleRate = config.sttSampleRate ?? DEFAULT_STT_SAMPLE_RATE;
+        this.ttsSampleRate = config.ttsSampleRate;
+    }
+    async startListening(language, handlers) {
+        if (typeof navigator === "undefined" || !navigator.mediaDevices) {
+            throw new Error("navigator.mediaDevices is unavailable for Sarvam STT");
+        }
+        if (this.sttSocket || this.sttMediaStream) {
+            await this.stopListening();
+        }
+        const generation = ++this.listeningGeneration;
+        const isCurrentSession = () => generation === this.listeningGeneration && this.sttSocket !== null;
+        const query = new URLSearchParams({
+            "language-code": language,
+            model: this.sttModel,
+            sample_rate: String(this.sttSampleRate),
+            input_audio_codec: "wav",
+            high_vad_sensitivity: "true",
+            vad_signals: "true",
+            flush_signal: "true",
+        });
+        const socket = new WebSocket(`${resolveWsUrl(this.sttUrl)}?${query.toString()}`);
+        this.sttSocket = socket;
+        socket.onmessage = (event) => {
+            if (!isCurrentSession()) {
+                return;
+            }
+            let message;
+            try {
+                message = JSON.parse(event.data);
+            }
+            catch {
+                return;
+            }
+            if (!message || typeof message !== "object") {
+                return;
+            }
+            switch (message.type) {
+                case "events": {
+                    const signalType = message.data?.signal_type;
+                    if (signalType === "START_SPEECH") {
+                        handlers.onSpeechStart?.();
+                    }
+                    else if (signalType === "END_SPEECH") {
+                        this.sendFlush(socket);
+                        handlers.onSpeechEnd?.();
+                    }
+                    break;
+                }
+                case "data": {
+                    const transcript = typeof message.data?.transcript === "string"
+                        ? message.data.transcript.trim()
+                        : "";
+                    if (!transcript) {
+                        return;
+                    }
+                    handlers.onPartial?.(transcript);
+                    handlers.onFinal?.(transcript);
+                    break;
+                }
+                case "transcript": {
+                    const transcript = typeof message.text === "string" ? message.text.trim() : "";
+                    if (!transcript) {
+                        return;
+                    }
+                    handlers.onPartial?.(transcript);
+                    handlers.onFinal?.(transcript);
+                    break;
+                }
+                case "error": {
+                    const errorMessage = typeof message.data?.error === "string"
+                        ? message.data.error
+                        : "Sarvam STT error";
+                    handlers.onError?.(new Error(errorMessage));
+                    break;
+                }
+                default:
+                    break;
+            }
+        };
+        socket.onerror = () => {
+            if (!isCurrentSession()) {
+                return;
+            }
+            handlers.onError?.(new Error("Sarvam STT socket error"));
+        };
+        await new Promise((resolve, reject) => {
+            socket.addEventListener("open", () => resolve(), { once: true });
+            socket.addEventListener("error", () => reject(new Error("Failed to connect to Sarvam STT relay")), { once: true });
+        });
+        if (!isCurrentSession()) {
+            return;
+        }
+        const mediaStream = await navigator.mediaDevices.getUserMedia({
+            audio: true,
+        });
+        const audioContext = new AudioContext();
+        const sourceNode = audioContext.createMediaStreamSource(mediaStream);
+        const processorNode = audioContext.createScriptProcessor(4096, 1, 1);
+        sourceNode.connect(processorNode);
+        processorNode.connect(audioContext.destination);
+        processorNode.onaudioprocess = (event) => {
+            if (!isCurrentSession()) {
+                return;
+            }
+            if (handlers.shouldSendAudio && !handlers.shouldSendAudio()) {
+                return;
+            }
+            if (socket.readyState !== WebSocket.OPEN) {
+                return;
+            }
+            const floatSamples = event.inputBuffer.getChannelData(0);
+            if (!floatSamples || floatSamples.length === 0) {
+                return;
+            }
+            try {
+                const pcm = downsampleTo16BitMono(floatSamples, audioContext.sampleRate, this.sttSampleRate);
+                const wav = encodeWav(pcm, this.sttSampleRate);
+                const audioBase64 = toBase64(wav);
+                socket.send(JSON.stringify({
+                    audio: {
+                        data: audioBase64,
+                        sample_rate: this.sttSampleRate,
+                        encoding: "audio/wav",
+                    },
+                }));
+            }
+            catch (error) {
+                handlers.onError?.(error instanceof Error
+                    ? error
+                    : new Error(`Failed to stream audio: ${String(error)}`));
+            }
+        };
+        this.sttContext = audioContext;
+        this.sttMediaStream = mediaStream;
+        this.sttSourceNode = sourceNode;
+        this.sttProcessorNode = processorNode;
+    }
+    sendFlush(socket) {
+        if (socket.readyState !== WebSocket.OPEN) {
+            return;
+        }
+        try {
+            socket.send(JSON.stringify({ type: "flush" }));
+        }
+        catch {
+        }
+    }
+    async stopListening() {
+        this.listeningGeneration++;
+        const socket = this.sttSocket;
+        this.sttSocket = null;
+        const processorNode = this.sttProcessorNode;
+        const sourceNode = this.sttSourceNode;
+        const audioContext = this.sttContext;
+        const mediaStream = this.sttMediaStream;
+        this.sttProcessorNode = null;
+        this.sttSourceNode = null;
+        this.sttContext = null;
+        this.sttMediaStream = null;
+        if (processorNode) {
+            try {
+                processorNode.disconnect();
+            }
+            catch {
+            }
+            processorNode.onaudioprocess = null;
+        }
+        if (sourceNode) {
+            try {
+                sourceNode.disconnect();
+            }
+            catch {
+            }
+        }
+        if (mediaStream) {
+            for (const track of mediaStream.getTracks()) {
+                try {
+                    track.stop();
+                }
+                catch {
+                }
+            }
+        }
+        if (audioContext) {
+            try {
+                await audioContext.close();
+            }
+            catch {
+            }
+        }
+        if (socket) {
+            this.sendFlush(socket);
+            try {
+                socket.onmessage = null;
+                socket.onerror = null;
+                socket.onclose = null;
+                socket.close();
+            }
+            catch {
+            }
+        }
+    }
+    async speak(text, opts) {
+        if (!text || !text.trim()) {
+            return;
+        }
+        this.stopSpeaking();
+        const generation = ++this.speechGeneration;
+        const language = opts?.language ?? "en-IN";
+        const body = {
+            text,
+            target_language_code: language,
+            model: this.ttsModel,
+            speaker: this.speaker,
+            output_audio_codec: "wav",
+        };
+        if (this.ttsSampleRate) {
+            body.speech_sample_rate = String(this.ttsSampleRate);
+        }
+        const response = await fetch(this.ttsUrl, {
+            method: "POST",
+            headers: {
+                "Content-Type": "application/json",
+            },
+            body: JSON.stringify(body),
+        });
+        if (!response.ok) {
+            const errorText = await response.text().catch(() => "");
+            throw new Error(`Sarvam TTS failed: ${response.status} ${response.statusText}${errorText ? ` - ${errorText}` : ""}`);
+        }
+        const payload = (await response.json());
+        const audioBase64 = Array.isArray(payload.audios) && payload.audios.length > 0
+            ? payload.audios[0]
+            : typeof payload.audio === "string"
+                ? payload.audio
+                : null;
+        if (!audioBase64) {
+            throw new Error("Sarvam TTS response does not contain audio data");
+        }
+        const binaryString = atob(audioBase64);
+        const bytes = new Uint8Array(binaryString.length);
+        for (let i = 0; i < binaryString.length; i++) {
+            bytes[i] = binaryString.charCodeAt(i);
+        }
+        if (generation !== this.speechGeneration) {
+            return;
+        }
+        const blob = new Blob([bytes], { type: "audio/wav" });
+        const blobUrl = URL.createObjectURL(blob);
+        if (generation !== this.speechGeneration) {
+            URL.revokeObjectURL(blobUrl);
+            return;
+        }
+        this.currentBlobUrl = blobUrl;
+        const audio = new Audio(blobUrl);
+        this.audioElement = audio;
+        await new Promise((resolve, reject) => {
+            audio.onended = () => {
+                this.cleanupAudioElement(audio, blobUrl);
+                resolve();
+            };
+            audio.onerror = () => {
+                this.cleanupAudioElement(audio, blobUrl);
+                reject(new Error("Sarvam TTS playback failed"));
+            };
+            audio
+                .play()
+                .then(() => {
+                if (generation === this.speechGeneration) {
+                    opts?.onAudioStart?.();
+                }
+            })
+                .catch((error) => {
+                this.cleanupAudioElement(audio, blobUrl);
+                reject(error instanceof Error ? error : new Error(String(error)));
+            });
+        });
+    }
+    stopSpeaking() {
+        this.speechGeneration++;
+        if (this.audioElement) {
+            try {
+                this.audioElement.pause();
+                this.audioElement.src = "";
+            }
+            catch {
+            }
+            this.audioElement = null;
+        }
+        if (this.currentBlobUrl) {
+            try {
+                URL.revokeObjectURL(this.currentBlobUrl);
+            }
+            catch {
+            }
+            this.currentBlobUrl = null;
+        }
+    }
+    async dispose() {
+        await this.stopListening();
+        this.stopSpeaking();
+    }
+    cleanupAudioElement(audio, blobUrl) {
+        if (this.audioElement === audio) {
+            this.audioElement = null;
+        }
+        if (this.currentBlobUrl === blobUrl) {
+            try {
+                URL.revokeObjectURL(blobUrl);
+            }
+            catch {
+            }
+            this.currentBlobUrl = null;
+        }
+    }
+}

package/dist/services/voice-feedback.d.ts

@@ -1,10 +1,13 @@
-import type { NavigationAction, ActionResult } from "../types.js";
+import type { NavigationAction, ActionResult, SpeechProvider } from "../types.js";
 import type { AzureSpeechService } from "./azure-speech.js";
 import type { BedrockService } from "./bedrock.js";
+import type { SarvamSpeechService } from "./sarvam-speech.js";
 interface VoiceFeedbackConfig {
     enabled?: boolean;
     language?: string;
-    azureSpeechService: AzureSpeechService;
+    speechProvider?: SpeechProvider;
+    azureSpeechService?: AzureSpeechService;
+    sarvamSpeechService?: SarvamSpeechService;
     bedrockService: BedrockService;
 }
 interface FeedbackContext {
@@ -21,7 +24,9 @@ interface FeedbackContext {
 export declare class VoiceFeedbackService {
     private enabled;
     private language;
+    private speechProvider;
     private azureSpeechService;
+    private sarvamSpeechService;
     private bedrockService;
     private synthesizer;
     private isSpeaking;

package/dist/services/voice-feedback.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"voice-feedback.d.ts","sourceRoot":"","sources":["../../src/services/voice-feedback.ts"],"names":[],"mappings":"AAMA,OAAO,KAAK,EAAE,gBAAgB,EAAE,YAAY,EAAE,MAAM,aAAa,CAAC;AAClE,OAAO,KAAK,EAAE,kBAAkB,EAAE,MAAM,mBAAmB,CAAC;AAC5D,OAAO,KAAK,EAAE,cAAc,EAAE,MAAM,cAAc,CAAC;AAEnD,UAAU,mBAAmB;IAC3B,OAAO,CAAC,EAAE,OAAO,CAAC;IAClB,QAAQ,CAAC,EAAE,MAAM,CAAC;IAClB,kBAAkB,EAAE,kBAAkB,CAAC;IACvC,cAAc,EAAE,cAAc,CAAC;CAChC;AAED,UAAU,eAAe;IACvB,MAAM,EAAE,gBAAgB,CAAC;IACzB,YAAY,CAAC,EAAE,YAAY,CAAC;IAC5B,UAAU,CAAC,EAAE,MAAM,CAAC;IACpB,WAAW,CAAC,EAAE;QACZ,GAAG,CAAC,EAAE,MAAM,CAAC;QACb,KAAK,CAAC,EAAE,MAAM,CAAC;QACf,IAAI,CAAC,EAAE,MAAM,CAAC;QACd,IAAI,CAAC,EAAE,MAAM,CAAC;KACf,CAAC;CACH;AAED,qBAAa,oBAAoB;IAC/B,OAAO,CAAC,OAAO,CAAU;IACzB,OAAO,CAAC,QAAQ,CAAS;IACzB,OAAO,CAAC,kBAAkB,CAAqB;IAC/C,OAAO,CAAC,cAAc,CAAiB;IACvC,OAAO,CAAC,WAAW,CAAsC;IACzD,OAAO,CAAC,UAAU,CAAkB;gBAExB,MAAM,EAAE,mBAAmB;IAiBvC,UAAU,CAAC,OAAO,EAAE,OAAO,GAAG,IAAI;IAclC,SAAS,IAAI,OAAO;IAOpB,WAAW,CAAC,QAAQ,EAAE,MAAM,GAAG,IAAI;YAQrB,qBAAqB;IAmCnC,OAAO,CAAC,mBAAmB;YAsBb,uBAAuB;IAsCrC,OAAO,CAAC,oBAAoB;IA4C5B,OAAO,CAAC,kBAAkB;YAgCZ,KAAK;IA0EnB,OAAO,CAAC,YAAY;IAkBd,eAAe,CAAC,OAAO,EAAE,eAAe,GAAG,OAAO,CAAC,IAAI,CAAC;IAoFxD,wBAAwB,CAAC,OAAO,EAAE,MAAM,GAAG,OAAO,CAAC,IAAI,CAAC;IAkBxD,mBAAmB,CACvB,OAAO,EAAE,WAAW,EACpB,SAAS,EAAE,MAAM,GAAG,UAAU,GAC7B,OAAO,CAAC,IAAI,CAAC;IAwBhB,OAAO,CAAC,kBAAkB;IAwF1B,OAAO,IAAI,IAAI;CAOhB"}
+{"version":3,"file":"voice-feedback.d.ts","sourceRoot":"","sources":["../../src/services/voice-feedback.ts"],"names":[],"mappings":"AAMA,OAAO,KAAK,EACV,gBAAgB,EAChB,YAAY,EACZ,cAAc,EACf,MAAM,aAAa,CAAC;AACrB,OAAO,KAAK,EAAE,kBAAkB,EAAE,MAAM,mBAAmB,CAAC;AAC5D,OAAO,KAAK,EAAE,cAAc,EAAE,MAAM,cAAc,CAAC;AACnD,OAAO,KAAK,EAAE,mBAAmB,EAAE,MAAM,oBAAoB,CAAC;AAI9D,UAAU,mBAAmB;IAC3B,OAAO,CAAC,EAAE,OAAO,CAAC;IAClB,QAAQ,CAAC,EAAE,MAAM,CAAC;IAClB,cAAc,CAAC,EAAE,cAAc,CAAC;IAChC,kBAAkB,CAAC,EAAE,kBAAkB,CAAC;IACxC,mBAAmB,CAAC,EAAE,mBAAmB,CAAC;IAC1C,cAAc,EAAE,cAAc,CAAC;CAChC;AAED,UAAU,eAAe;IACvB,MAAM,EAAE,gBAAgB,CAAC;IACzB,YAAY,CAAC,EAAE,YAAY,CAAC;IAC5B,UAAU,CAAC,EAAE,MAAM,CAAC;IACpB,WAAW,CAAC,EAAE;QACZ,GAAG,CAAC,EAAE,MAAM,CAAC;QACb,KAAK,CAAC,EAAE,MAAM,CAAC;QACf,IAAI,CAAC,EAAE,MAAM,CAAC;QACd,IAAI,CAAC,EAAE,MAAM,CAAC;KACf,CAAC;CACH;AAED,qBAAa,oBAAoB;IAC/B,OAAO,CAAC,OAAO,CAAU;IACzB,OAAO,CAAC,QAAQ,CAAS;IACzB,OAAO,CAAC,cAAc,CAAiB;IACvC,OAAO,CAAC,kBAAkB,CAA4B;IACtD,OAAO,CAAC,mBAAmB,CAA6B;IACxD,OAAO,CAAC,cAAc,CAAiB;IACvC,OAAO,CAAC,WAAW,CAAsC;IACzD,OAAO,CAAC,UAAU,CAAkB;gBAExB,MAAM,EAAE,mBAAmB;IA2BvC,UAAU,CAAC,OAAO,EAAE,OAAO,GAAG,IAAI;IAclC,SAAS,IAAI,OAAO;IAOpB,WAAW,CAAC,QAAQ,EAAE,MAAM,GAAG,IAAI;YAQrB,qBAAqB;IA2CnC,OAAO,CAAC,mBAAmB;YAsBb,uBAAuB;IAwCrC,OAAO,CAAC,oBAAoB;IA4C5B,OAAO,CAAC,kBAAkB;YAuCZ,KAAK;IAyFnB,OAAO,CAAC,YAAY;IAwBd,eAAe,CAAC,OAAO,EAAE,eAAe,GAAG,OAAO,CAAC,IAAI,CAAC;IAoFxD,wBAAwB,CAAC,OAAO,EAAE,MAAM,GAAG,OAAO,CAAC,IAAI,CAAC;IAkBxD,mBAAmB,CACvB,OAAO,EAAE,WAAW,EACpB,SAAS,EAAE,MAAM,GAAG,UAAU,GAC7B,OAAO,CAAC,IAAI,CAAC;IAwBhB,OAAO,CAAC,kBAAkB;IAwF1B,OAAO,IAAI,IAAI;CAWhB"}