@openclaw/voice-call 2026.1.29 → 2026.2.1

package/src/media-stream.ts

@@ -9,9 +9,7 @@
 
 import type { IncomingMessage } from "node:http";
 import type { Duplex } from "node:stream";
-
 import { WebSocket, WebSocketServer } from "ws";
-
 import type {
   OpenAIRealtimeSTTProvider,
   RealtimeSTTSession,
@@ -87,10 +85,7 @@ export class MediaStreamHandler {
   /**
    * Handle new WebSocket connection from Twilio.
    */
-  private async handleConnection(
-    ws: WebSocket,
-    _request: IncomingMessage,
-  ): Promise<void> {
+  private async handleConnection(ws: WebSocket, _request: IncomingMessage): Promise<void> {
     let session: StreamSession | null = null;
 
     ws.on("message", async (data: Buffer) => {
@@ -140,16 +135,11 @@ export class MediaStreamHandler {
   /**
    * Handle stream start event.
    */
-  private async handleStart(
-    ws: WebSocket,
-    message: TwilioMediaMessage,
-  ): Promise<StreamSession> {
+  private async handleStart(ws: WebSocket, message: TwilioMediaMessage): Promise<StreamSession> {
     const streamSid = message.streamSid || "";
     const callSid = message.start?.callSid || "";
 
-    console.log(
-      `[MediaStream] Stream started: ${streamSid} (call: ${callSid})`,
-    );
+    console.log(`[MediaStream] Stream started: ${streamSid} (call: ${callSid})`);
 
     // Create STT session
     const sttSession = this.config.sttProvider.createSession();
@@ -181,10 +171,7 @@ export class MediaStreamHandler {
 
     // Connect to OpenAI STT (non-blocking, log errors but don't fail the call)
     sttSession.connect().catch((err) => {
-      console.warn(
-        `[MediaStream] STT connection failed (TTS still works):`,
-        err.message,
-      );
+      console.warn(`[MediaStream] STT connection failed (TTS still works):`, err.message);
     });
 
     return session;
@@ -252,10 +239,7 @@ export class MediaStreamHandler {
    * Queue a TTS operation for sequential playback.
    * Only one TTS operation plays at a time per stream to prevent overlap.
    */
-  async queueTts(
-    streamSid: string,
-    playFn: (signal: AbortSignal) => Promise<void>,
-  ): Promise<void> {
+  async queueTts(streamSid: string, playFn: (signal: AbortSignal) => Promise<void>): Promise<void> {
     const queue = this.getTtsQueue(streamSid);
     let resolveEntry: () => void;
     let rejectEntry: (error: unknown) => void;
@@ -292,9 +276,7 @@ export class MediaStreamHandler {
    * Get active session by call ID.
    */
   getSessionByCallId(callId: string): StreamSession | undefined {
-    return [...this.sessions.values()].find(
-      (session) => session.callId === callId,
-    );
+    return [...this.sessions.values()].find((session) => session.callId === callId);
   }
 
   /**
@@ -311,7 +293,9 @@ export class MediaStreamHandler {
 
   private getTtsQueue(streamSid: string): TtsQueueEntry[] {
     const existing = this.ttsQueues.get(streamSid);
-    if (existing) return existing;
+    if (existing) {
+      return existing;
+    }
     const queue: TtsQueueEntry[] = [];
     this.ttsQueues.set(streamSid, queue);
     return queue;
@@ -355,7 +339,9 @@ export class MediaStreamHandler {
 
   private clearTtsState(streamSid: string): void {
     const queue = this.ttsQueues.get(streamSid);
-    if (queue) queue.length = 0;
+    if (queue) {
+      queue.length = 0;
+    }
     this.ttsActiveControllers.get(streamSid)?.abort();
     this.ttsActiveControllers.delete(streamSid);
     this.ttsPlaying.delete(streamSid);

package/src/providers/mock.ts

@@ -1,5 +1,4 @@
 import crypto from "node:crypto";
-
 import type {
   EndReason,
   HangupCallInput,
@@ -37,11 +36,15 @@ export class MockProvider implements VoiceCallProvider {
       if (Array.isArray(payload.events)) {
         for (const evt of payload.events) {
           const normalized = this.normalizeEvent(evt);
-          if (normalized) events.push(normalized);
+          if (normalized) {
+            events.push(normalized);
+          }
         }
       } else if (payload.event) {
         const normalized = this.normalizeEvent(payload.event);
-        if (normalized) events.push(normalized);
+        if (normalized) {
+          events.push(normalized);
+        }
       }
 
       return { events, statusCode: 200 };
@@ -50,10 +53,10 @@ export class MockProvider implements VoiceCallProvider {
     }
   }
 
-  private normalizeEvent(
-    evt: Partial<NormalizedEvent>,
-  ): NormalizedEvent | null {
-    if (!evt.type || !evt.callId) return null;
+  private normalizeEvent(evt: Partial<NormalizedEvent>): NormalizedEvent | null {
+    if (!evt.type || !evt.callId) {
+      return null;
+    }
 
     const base = {
       id: evt.id || crypto.randomUUID(),
@@ -96,9 +99,7 @@ export class MockProvider implements VoiceCallProvider {
       }
 
       case "call.silence": {
-        const payload = evt as Partial<
-          NormalizedEvent & { durationMs?: number }
-        >;
+        const payload = evt as Partial<NormalizedEvent & { durationMs?: number }>;
         return {
           ...base,
           type: evt.type,
@@ -116,9 +117,7 @@ export class MockProvider implements VoiceCallProvider {
       }
 
       case "call.ended": {
-        const payload = evt as Partial<
-          NormalizedEvent & { reason?: EndReason }
-        >;
+        const payload = evt as Partial<NormalizedEvent & { reason?: EndReason }>;
         return {
           ...base,
           type: evt.type,
@@ -127,9 +126,7 @@ export class MockProvider implements VoiceCallProvider {
       }
 
       case "call.error": {
-        const payload = evt as Partial<
-          NormalizedEvent & { error?: string; retryable?: boolean }
-        >;
+        const payload = evt as Partial<NormalizedEvent & { error?: string; retryable?: boolean }>;
         return {
           ...base,
           type: evt.type,

package/src/providers/plivo.test.ts

@@ -1,5 +1,4 @@
 import { describe, expect, it } from "vitest";
-
 import { PlivoProvider } from "./plivo.js";
 
 describe("PlivoProvider", () => {

package/src/providers/plivo.ts

@@ -1,5 +1,4 @@
 import crypto from "node:crypto";
-
 import type { PlivoConfig } from "../config.js";
 import type {
   HangupCallInput,
@@ -13,9 +12,9 @@ import type {
   WebhookContext,
   WebhookVerificationResult,
 } from "../types.js";
+import type { VoiceCallProvider } from "./base.js";
 import { escapeXml } from "../voice-mapping.js";
 import { reconstructWebhookUrl, verifyPlivoWebhook } from "../webhook-security.js";
-import type { VoiceCallProvider } from "./base.js";
 
 export interface PlivoProviderOptions {
   /** Override public URL origin for signature verification */
@@ -103,8 +102,7 @@ export class PlivoProvider implements VoiceCallProvider {
   }
 
   parseWebhookEvent(ctx: WebhookContext): ProviderWebhookParseResult {
-    const flow =
-      typeof ctx.query?.flow === "string" ? ctx.query.flow.trim() : "";
+    const flow = typeof ctx.query?.flow === "string" ? ctx.query.flow.trim() : "";
 
     const parsed = this.parseBody(ctx.rawBody);
     if (!parsed) {
@@ -124,7 +122,9 @@ export class PlivoProvider implements VoiceCallProvider {
     if (flow === "xml-speak") {
       const callId = this.getCallIdFromQuery(ctx);
       const pending = callId ? this.pendingSpeakByCallId.get(callId) : undefined;
-      if (callId) this.pendingSpeakByCallId.delete(callId);
+      if (callId) {
+        this.pendingSpeakByCallId.delete(callId);
+      }
 
       const xml = pending
         ? PlivoProvider.xmlSpeak(pending.text, pending.locale)
@@ -139,10 +139,10 @@ export class PlivoProvider implements VoiceCallProvider {
 
     if (flow === "xml-listen") {
       const callId = this.getCallIdFromQuery(ctx);
-      const pending = callId
-        ? this.pendingListenByCallId.get(callId)
-        : undefined;
-      if (callId) this.pendingListenByCallId.delete(callId);
+      const pending = callId ? this.pendingListenByCallId.get(callId) : undefined;
+      if (callId) {
+        this.pendingListenByCallId.delete(callId);
+      }
 
       const actionUrl = this.buildActionUrl(ctx, {
         flow: "getinput",
@@ -180,10 +180,7 @@ export class PlivoProvider implements VoiceCallProvider {
     };
   }
 
-  private normalizeEvent(
-    params: URLSearchParams,
-    callIdOverride?: string,
-  ): NormalizedEvent | null {
+  private normalizeEvent(params: URLSearchParams, callIdOverride?: string): NormalizedEvent | null {
     const callUuid = params.get("CallUUID") || "";
     const requestUuid = params.get("RequestUUID") || "";
 
@@ -329,11 +326,9 @@ export class PlivoProvider implements VoiceCallProvider {
   }
 
   async playTts(input: PlayTtsInput): Promise<void> {
-    const callUuid = this.requestUuidToCallUuid.get(input.providerCallId) ??
-      input.providerCallId;
+    const callUuid = this.requestUuidToCallUuid.get(input.providerCallId) ?? input.providerCallId;
     const webhookBase =
-      this.callUuidToWebhookUrl.get(callUuid) ||
-      this.callIdToWebhookUrl.get(input.callId);
+      this.callUuidToWebhookUrl.get(callUuid) || this.callIdToWebhookUrl.get(input.callId);
     if (!webhookBase) {
       throw new Error("Missing webhook URL for this call (provider state missing)");
     }
@@ -364,11 +359,9 @@ export class PlivoProvider implements VoiceCallProvider {
   }
 
   async startListening(input: StartListeningInput): Promise<void> {
-    const callUuid = this.requestUuidToCallUuid.get(input.providerCallId) ??
-      input.providerCallId;
+    const callUuid = this.requestUuidToCallUuid.get(input.providerCallId) ?? input.providerCallId;
     const webhookBase =
-      this.callUuidToWebhookUrl.get(callUuid) ||
-      this.callIdToWebhookUrl.get(input.callId);
+      this.callUuidToWebhookUrl.get(callUuid) || this.callIdToWebhookUrl.get(input.callId);
     if (!webhookBase) {
       throw new Error("Missing webhook URL for this call (provider state missing)");
     }
@@ -403,7 +396,9 @@ export class PlivoProvider implements VoiceCallProvider {
 
   private static normalizeNumber(numberOrSip: string): string {
     const trimmed = numberOrSip.trim();
-    if (trimmed.toLowerCase().startsWith("sip:")) return trimmed;
+    if (trimmed.toLowerCase().startsWith("sip:")) {
+      return trimmed;
+    }
     return trimmed.replace(/[^\d+]/g, "");
   }
 
@@ -427,10 +422,7 @@ export class PlivoProvider implements VoiceCallProvider {
 </Response>`;
   }
 
-  private static xmlGetInputSpeech(params: {
-    actionUrl: string;
-    language?: string;
-  }): string {
+  private static xmlGetInputSpeech(params: { actionUrl: string; language?: string }): string {
     const language = params.language || "en-US";
     return `<?xml version="1.0" encoding="UTF-8"?>
 <Response>
@@ -453,12 +445,16 @@ export class PlivoProvider implements VoiceCallProvider {
     opts: { flow: string; callId?: string },
   ): string | null {
     const base = PlivoProvider.baseWebhookUrlFromCtx(ctx);
-    if (!base) return null;
+    if (!base) {
+      return null;
+    }
 
     const u = new URL(base);
     u.searchParams.set("provider", "plivo");
     u.searchParams.set("flow", opts.flow);
-    if (opts.callId) u.searchParams.set("callId", opts.callId);
+    if (opts.callId) {
+      u.searchParams.set("callId", opts.callId);
+    }
     return u.toString();
   }
 
@@ -491,7 +487,9 @@ export class PlivoProvider implements VoiceCallProvider {
 
     for (const key of candidates) {
       const value = params.get(key);
-      if (value && value.trim()) return value.trim();
+      if (value && value.trim()) {
+        return value.trim();
+      }
     }
     return null;
   }

package/src/providers/stt-openai-realtime.ts

@@ -155,7 +155,9 @@ class OpenAIRealtimeSTTSession implements RealtimeSTTSession {
 
       this.ws.on("error", (error) => {
         console.error("[RealtimeSTT] WebSocket error:", error);
-        if (!this.connected) reject(error);
+        if (!this.connected) {
+          reject(error);
+        }
       });
 
       this.ws.on("close", (code, reason) => {
@@ -183,9 +185,7 @@ class OpenAIRealtimeSTTSession implements RealtimeSTTSession {
       return;
     }
 
-    if (
-      this.reconnectAttempts >= OpenAIRealtimeSTTSession.MAX_RECONNECT_ATTEMPTS
-    ) {
+    if (this.reconnectAttempts >= OpenAIRealtimeSTTSession.MAX_RECONNECT_ATTEMPTS) {
       console.error(
         `[RealtimeSTT] Max reconnect attempts (${OpenAIRealtimeSTTSession.MAX_RECONNECT_ATTEMPTS}) reached`,
       );
@@ -193,9 +193,7 @@ class OpenAIRealtimeSTTSession implements RealtimeSTTSession {
     }
 
     this.reconnectAttempts++;
-    const delay =
-      OpenAIRealtimeSTTSession.RECONNECT_DELAY_MS *
-      2 ** (this.reconnectAttempts - 1);
+    const delay = OpenAIRealtimeSTTSession.RECONNECT_DELAY_MS * 2 ** (this.reconnectAttempts - 1);
     console.log(
       `[RealtimeSTT] Reconnecting ${this.reconnectAttempts}/${OpenAIRealtimeSTTSession.MAX_RECONNECT_ATTEMPTS} in ${delay}ms...`,
     );
@@ -262,7 +260,9 @@ class OpenAIRealtimeSTTSession implements RealtimeSTTSession {
   }
 
   sendAudio(muLawData: Buffer): void {
-    if (!this.connected) return;
+    if (!this.connected) {
+      return;
+    }
     this.sendEvent({
       type: "input_audio_buffer.append",
       audio: muLawData.toString("base64"),

package/src/providers/telnyx.ts

@@ -1,5 +1,4 @@
 import crypto from "node:crypto";
-
 import type { TelnyxConfig } from "../config.js";
 import type {
   EndReason,
@@ -161,9 +160,7 @@ export class TelnyxProvider implements VoiceCallProvider {
     let callId = "";
     if (data.payload?.client_state) {
       try {
-        callId = Buffer.from(data.payload.client_state, "base64").toString(
-          "utf8",
-        );
+        callId = Buffer.from(data.payload.client_state, "base64").toString("utf8");
       } catch {
         // Fallback if not valid Base64
         callId = data.payload.client_state;
@@ -312,13 +309,10 @@ export class TelnyxProvider implements VoiceCallProvider {
    * Start transcription (STT) via Telnyx.
    */
   async startListening(input: StartListeningInput): Promise<void> {
-    await this.apiRequest(
-      `/calls/${input.providerCallId}/actions/transcription_start`,
-      {
-        command_id: crypto.randomUUID(),
-        language: input.language || "en",
-      },
-    );
+    await this.apiRequest(`/calls/${input.providerCallId}/actions/transcription_start`, {
+      command_id: crypto.randomUUID(),
+      language: input.language || "en",
+    });
   }
 
   /**

package/src/providers/tts-openai.ts

@@ -84,9 +84,7 @@ export class OpenAITTSProvider {
     this.instructions = config.instructions;
 
     if (!this.apiKey) {
-      throw new Error(
-        "OpenAI API key required (set OPENAI_API_KEY or pass apiKey)",
-      );
+      throw new Error("OpenAI API key required (set OPENAI_API_KEY or pass apiKey)");
     }
   }
 
@@ -207,19 +205,19 @@ function linearToMulaw(sample: number): number {
 
   // Get sign bit
   const sign = sample < 0 ? 0x80 : 0;
-  if (sample < 0) sample = -sample;
+  if (sample < 0) {
+    sample = -sample;
+  }
 
   // Clip to prevent overflow
-  if (sample > CLIP) sample = CLIP;
+  if (sample > CLIP) {
+    sample = CLIP;
+  }
 
   // Add bias and find segment
   sample += BIAS;
   let exponent = 7;
-  for (
-    let expMask = 0x4000;
-    (sample & expMask) === 0 && exponent > 0;
-    exponent--, expMask >>= 1
-  ) {
+  for (let expMask = 0x4000; (sample & expMask) === 0 && exponent > 0; exponent--, expMask >>= 1) {
     // Find the segment (exponent)
   }
 
@@ -252,10 +250,7 @@ export function mulawToLinear(mulaw: number): number {
  * Chunk audio buffer into 20ms frames for streaming.
  * At 8kHz mono, 20ms = 160 samples = 160 bytes (mu-law).
  */
-export function chunkAudio(
-  audio: Buffer,
-  chunkSize = 160,
-): Generator<Buffer, void, unknown> {
+export function chunkAudio(audio: Buffer, chunkSize = 160): Generator<Buffer, void, unknown> {
   return (function* () {
     for (let i = 0; i < audio.length; i += chunkSize) {
       yield audio.subarray(i, Math.min(i + chunkSize, audio.length));

package/src/providers/twilio/api.ts

@@ -9,19 +9,16 @@ export async function twilioApiRequest<T = unknown>(params: {
   const bodyParams =
     params.body instanceof URLSearchParams
       ? params.body
-      : Object.entries(params.body).reduce<URLSearchParams>(
-          (acc, [key, value]) => {
-            if (Array.isArray(value)) {
-              for (const entry of value) {
-                acc.append(key, entry);
-              }
-            } else if (typeof value === "string") {
-              acc.append(key, value);
+      : Object.entries(params.body).reduce<URLSearchParams>((acc, [key, value]) => {
+          if (Array.isArray(value)) {
+            for (const entry of value) {
+              acc.append(key, entry);
             }
-            return acc;
-          },
-          new URLSearchParams(),
-        );
+          } else if (typeof value === "string") {
+            acc.append(key, value);
+          }
+          return acc;
+        }, new URLSearchParams());
 
   const response = await fetch(`${params.baseUrl}${params.endpoint}`, {
     method: "POST",

package/src/providers/twilio/webhook.ts

@@ -1,7 +1,6 @@
 import type { WebhookContext, WebhookVerificationResult } from "../../types.js";
-import { verifyTwilioWebhook } from "../../webhook-security.js";
-
 import type { TwilioProviderOptions } from "../twilio.js";
+import { verifyTwilioWebhook } from "../../webhook-security.js";
 
 export function verifyTwilioProviderWebhook(params: {
   ctx: WebhookContext;
@@ -11,8 +10,7 @@ export function verifyTwilioProviderWebhook(params: {
 }): WebhookVerificationResult {
   const result = verifyTwilioWebhook(params.ctx, params.authToken, {
     publicUrl: params.currentPublicUrl || undefined,
-    allowNgrokFreeTierLoopbackBypass:
-      params.options.allowNgrokFreeTierLoopbackBypass ?? false,
+    allowNgrokFreeTierLoopbackBypass: params.options.allowNgrokFreeTierLoopbackBypass ?? false,
     skipVerification: params.options.skipVerification,
   });
 

package/src/providers/twilio.test.ts

@@ -1,5 +1,4 @@
 import { describe, expect, it } from "vitest";
-
 import type { WebhookContext } from "../types.js";
 import { TwilioProvider } from "./twilio.js";
 
@@ -12,10 +11,7 @@ function createProvider(): TwilioProvider {
   );
 }
 
-function createContext(
-  rawBody: string,
-  query?: WebhookContext["query"],
-): WebhookContext {
+function createContext(rawBody: string, query?: WebhookContext["query"]): WebhookContext {
   return {
     headers: {},
     rawBody,