From 36f129089b3115ebde8a11f7cf82ef4831beadcf Mon Sep 17 00:00:00 2001
From: Danny Avila <danny@librechat.ai>
Date: Sun, 30 Aug 2026 06:54:16 -0400
Subject: [PATCH] =?UTF-8?q?=F0=9F=94=8A=20fix:=20Autoplay=20Latest=20Messa?=
 =?UTF-8?q?ge=20for=20Browser=20TTS=20and=20short=20responses=20(#15347)?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

* 🔊 fix: Autoplay Latest Message for Browser TTS and short responses

Automatic playback never reached the Web Speech API and dropped audio for
short responses. Two independent defects:

Autoplay ignored the selected TTS engine. `ChatForm` mounted `StreamAudio`
whenever automatic playback was on, and that component always POSTs to
`/api/files/speech/tts`, so the Browser engine silently fell through to the
external endpoint and `speechSynthesis.speak()` was never called. The
engine-aware `useTextToSpeech` hook had no consumers. Autoplay now selects
its driver by engine the same way `MessageAudio` does, with the gate shared
between both drivers so they trigger on identical terms.

`MediaSourceAppender` stranded its queue. `tryAppendNextChunk` only ran from
`addData` and `updateend`, while `sourceopen` — which fires only once a media
element attaches the object URL — created the `SourceBuffer` without draining
what had queued up behind it. A response read in full before that event left
every chunk in the queue with no `updateend` to ever restart the drain, so the
element sat at `readyState: 0` and never played. Short responses lost the race;
longer ones streamed past `sourceopen` and worked. The handler now drains, and
`close()` (previously never called) ends the stream once the queue empties.

Also in the same path: the error branch had its timeout comparison inverted, so
real failures were logged as timeouts and swallowed; and the non-MSE fallback
that hands the element a finished blob lived inside the cache-only branch,
leaving browsers without MSE support for `audio/mpeg` with no audio at all
unless Cache TTS happened to be enabled.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0184N7d9kgpdqmbCb9ujMmNX

* 🧹 style: Sort imports in autoplay files

`npm run sort-imports:check` on the changed files.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_0184N7d9kgpdqmbCb9ujMmNX

* 🔇 fix: End the media source when no audio ever arrives

Codex P2 on #15347. The `hasAppended` guard in `tryEndOfStream` was defensive
against a throw that cannot happen: `endOfStream()` only raises
`InvalidStateError` while the source is not open or a buffer is updating, both
of which the remaining guards already cover. Nothing about an empty buffer
throws.

The guard's actual effect was to strand the two cases it was meant to protect.
A TTS response that completes with zero bytes, or a read timeout that fires
before the first byte, both reach `close()` with the MediaSource URL already
attached to the element — and left it `open` forever, so the element waited on
a source that could never receive data and kept its streaming resources
---
 .../components/Chat/Input/AutoPlayAudio.tsx   |  34 ++++
 .../components/Chat/Input/BrowserAudio.tsx    |  57 ++++++
 client/src/components/Chat/Input/ChatForm.tsx |   4 +-
 .../src/components/Chat/Input/StreamAudio.tsx |  85 ++++-----
 .../Input/__tests__/AutoPlayAudio.spec.tsx    |  56 ++++++
 .../Input/__tests__/BrowserAudio.spec.tsx     | 149 ++++++++++++++++
 client/src/hooks/Audio/MediaSourceAppender.ts |  53 +++++-
 .../__tests__/MediaSourceAppender.spec.ts     | 164 ++++++++++++++++++
 client/src/hooks/Audio/index.ts               |   1 +
 client/src/hooks/Audio/useAutoplayTrigger.ts  |  37 ++++
 .../src/hooks/Input/useTextToSpeechBrowser.ts |  75 ++++----
 11 files changed, 622 insertions(+), 93 deletions(-)
 create mode 100644 client/src/components/Chat/Input/AutoPlayAudio.tsx
 create mode 100644 client/src/components/Chat/Input/BrowserAudio.tsx
 create mode 100644 client/src/components/Chat/Input/__tests__/AutoPlayAudio.spec.tsx
 create mode 100644 client/src/components/Chat/Input/__tests__/BrowserAudio.spec.tsx
 create mode 100644 client/src/hooks/Audio/__tests__/MediaSourceAppender.spec.ts
 create mode 100644 client/src/hooks/Audio/useAutoplayTrigger.ts

diff --git a/client/src/components/Chat/Input/AutoPlayAudio.tsx b/client/src/components/Chat/Input/AutoPlayAudio.tsx
new file mode 100644
index 0000000000..7cc5e4a66b
--- /dev/null
+++ b/client/src/components/Chat/Input/AutoPlayAudio.tsx
@@ -0,0 +1,34 @@
+import { memo } from 'react';
+import { useRecoilValue } from 'recoil';
+import BrowserAudio from './BrowserAudio';
+import StreamAudio from './StreamAudio';
+import { TTSEndpoints } from '~/common';
+import store from '~/store';
+
+/**
+ * Mount point for "Autoplay Latest Message". Playback has to honor the selected TTS engine
+ * the same way the message speaker button does, otherwise the Browser engine silently falls
+ * through to the external endpoint and never reaches the Web Speech API.
+ */
+function AutoPlayAudio({ index = 0 }) {
+  const engineTTS = useRecoilValue<string>(store.engineTTS);
+  const speechSettingsInitialized = useRecoilValue(store.speechSettingsInitialized);
+
+  if (!speechSettingsInitialized) {
+    return null;
+  }
+
+  const AutoPlayComponents = {
+    [TTSEndpoints.browser]: BrowserAudio,
+    [TTSEndpoints.external]: StreamAudio,
+  };
+
+  const SelectedAutoPlay = AutoPlayComponents[engineTTS];
+  if (!SelectedAutoPlay) {
+    return null;
+  }
+
+  return <SelectedAutoPlay index={index} />;
+}
+
+export default memo(AutoPlayAudio);
diff --git a/client/src/components/Chat/Input/BrowserAudio.tsx b/client/src/components/Chat/Input/BrowserAudio.tsx
new file mode 100644
index 0000000000..262310f7b2
--- /dev/null
+++ b/client/src/components/Chat/Input/BrowserAudio.tsx
@@ -0,0 +1,57 @@
+import { useEffect } from 'react';
+import { useSetRecoilState } from 'recoil';
+import { useParams } from 'react-router-dom';
+import { parseTextParts } from 'librechat-data-provider';
+import useTextToSpeechBrowser from '~/hooks/Input/useTextToSpeechBrowser';
+import useAutoplayTrigger from '~/hooks/Audio/useAutoplayTrigger';
+import { logger } from '~/utils';
+import store from '~/store';
+
+/**
+ * Autoplay driver for the Browser TTS engine. `StreamAudio` covers the external engine by
+ * streaming server audio into the global `<audio>` element; the Web Speech API has no such
+ * element, so the utterance is spoken directly off the shared autoplay gate.
+ */
+export default function BrowserAudio({ index = 0 }) {
+  const { conversationId: paramId } = useParams();
+  const setIsSpeaking = useSetRecoilState(store.globalAudioPlayingFamily(index));
+  const setAudioRunId = useSetRecoilState(store.audioRunFamily(index));
+
+  const { shouldPlay, activeRunId, latestMessage } = useAutoplayTrigger(index);
+  const { generateSpeechLocal, cancelSpeechLocal } = useTextToSpeechBrowser({ setIsSpeaking });
+
+  /**
+   * Leaving a conversation (or unmounting) stops whatever is still being spoken. React runs
+   * every cleanup before any effect body, so the utterance the effect below is about to start
+   * survives the `/c/new` -> `/c/:id` navigation that lands in the same commit as finalization.
+   */
+  useEffect(() => {
+    return () => cancelSpeechLocal();
+    // We only want the effect to run when the paramId changes
+    // eslint-disable-next-line react-hooks/exhaustive-deps
+  }, [paramId]);
+
+  useEffect(() => {
+    if (!shouldPlay || activeRunId == null || latestMessage == null) {
+      return;
+    }
+
+    const text =
+      Array.isArray(latestMessage.content) && latestMessage.content.length > 0
+        ? parseTextParts(latestMessage.content)
+        : (latestMessage.text ?? '');
+
+    if (!text) {
+      return;
+    }
+
+    logger.log('BrowserAudio.tsx - speaking message:', latestMessage.messageId);
+    /** Only claim the run once an utterance was queued, so a not-yet-loaded voice list
+     *  retries on the next render instead of silently swallowing the playback. */
+    if (generateSpeechLocal(text)) {
+      setAudioRunId(activeRunId);
+    }
+  }, [shouldPlay, activeRunId, latestMessage, setAudioRunId, generateSpeechLocal]);
+
+  return null;
+}
diff --git a/client/src/components/Chat/Input/ChatForm.tsx b/client/src/components/Chat/Input/ChatForm.tsx
index 0bfe3f7416..0a1266216a 100644
--- a/client/src/components/Chat/Input/ChatForm.tsx
+++ b/client/src/components/Chat/Input/ChatForm.tsx
@@ -51,9 +51,9 @@ import TextareaHeader from './TextareaHeader';
 import PromptsCommand from './PromptsCommand';
 import SkillsCommand from './SkillsCommand';
 import AudioRecorder from './AudioRecorder';
+import AutoPlayAudio from './AutoPlayAudio';
 import CollapseChat from './CollapseChat';
 import QuoteButton from './QuoteButton';
-import StreamAudio from './StreamAudio';
 import TokenUsage from './TokenUsage';
 import StopButton from './StopButton';
 import SendButton from './SendButton';
@@ -769,7 +769,7 @@ const ChatForm = memo(function ChatForm({
                       )}
                 </div>
               </div>
-              {TextToSpeech && automaticPlayback && <StreamAudio index={index} />}
+              {TextToSpeech && automaticPlayback && <AutoPlayAudio index={index} />}
             </div>
           </div>
         </div>
diff --git a/client/src/components/Chat/Input/StreamAudio.tsx b/client/src/components/Chat/Input/StreamAudio.tsx
index 09011b27ab..d7e788da46 100644
--- a/client/src/components/Chat/Input/StreamAudio.tsx
+++ b/client/src/components/Chat/Input/StreamAudio.tsx
@@ -4,11 +4,15 @@ import { QueryKeys } from 'librechat-data-provider';
 import { useQueryClient } from '@tanstack/react-query';
 import { useRecoilState, useRecoilValue, useSetRecoilState } from 'recoil';
 import type { TMessage } from 'librechat-data-provider';
-import { useCustomAudioRef, MediaSourceAppender, usePauseGlobalAudio } from '~/hooks/Audio';
-import { useLatestMessage } from '~/hooks/Messages/useLatestMessage';
-import { getLatestText, logger } from '~/utils';
+import {
+  useCustomAudioRef,
+  useAutoplayTrigger,
+  MediaSourceAppender,
+  usePauseGlobalAudio,
+} from '~/hooks/Audio';
 import { useAuthContext } from '~/hooks';
 import { globalAudioId } from '~/common';
+import { logger } from '~/utils';
 import store from '~/store';
 
 function timeoutPromise(ms: number, message?: string) {
@@ -27,15 +31,13 @@ export default function StreamAudio({ index = 0 }) {
   const playbackRate = useRecoilValue(store.playbackRate);
 
   const voice = useRecoilValue(store.voice);
-  const activeRunId = useRecoilValue(store.activeRunFamily(index));
   const automaticPlayback = useRecoilValue(store.automaticPlayback);
-  const isSubmitting = useRecoilValue(store.isSubmittingFamily(index));
-  const latestMessage = useLatestMessage(index);
   const setIsPlaying = useSetRecoilState(store.globalAudioPlayingFamily(index));
-  const [audioRunId, setAudioRunId] = useRecoilState(store.audioRunFamily(index));
+  const setAudioRunId = useSetRecoilState(store.audioRunFamily(index));
   const [isFetching, setIsFetching] = useRecoilState(store.globalAudioFetchingFamily(index));
   const [globalAudioURL, setGlobalAudioURL] = useRecoilState(store.globalAudioURLFamily(index));
 
+  const { shouldPlay, activeRunId, latestMessage } = useAutoplayTrigger(index);
   const { audioRef } = useCustomAudioRef({ setIsPlaying });
   const { pauseGlobalAudio } = usePauseGlobalAudio();
 
@@ -49,28 +51,13 @@ export default function StreamAudio({ index = 0 }) {
   );
 
   useEffect(() => {
-    const latestText = getLatestText(latestMessage);
-
-    const shouldFetch = !!(
-      token != null &&
-      automaticPlayback &&
-      !isSubmitting &&
-      latestMessage &&
-      !latestMessage.isCreatedByUser &&
-      latestText &&
-      latestMessage.messageId &&
-      !latestMessage.messageId.includes('_') &&
-      !isFetching &&
-      activeRunId != null &&
-      activeRunId !== audioRunId
-    );
-
-    if (!shouldFetch) {
+    if (!(token != null && automaticPlayback && shouldPlay && !isFetching)) {
       return;
     }
 
     async function fetchAudio() {
       setIsFetching(true);
+      let mediaSource: MediaSourceAppender | undefined;
 
       try {
         if (audioRef.current) {
@@ -112,12 +99,15 @@ export default function StreamAudio({ index = 0 }) {
         const type = 'audio/mpeg';
         const browserSupportsType =
           typeof MediaSource !== 'undefined' && MediaSource.isTypeSupported(type);
-        let mediaSource: MediaSourceAppender | undefined;
         if (browserSupportsType) {
           mediaSource = new MediaSourceAppender(type);
           setGlobalAudioURL(mediaSource.mediaSourceUrl);
         }
 
+        /** Browsers without MSE support for the type (Safari, for one) can only be handed the
+         *  audio once it is fully read, so that path always buffers — caching opts MSE in too. */
+        const shouldBufferChunks = cacheTTS || !browserSupportsType;
+
         let done = false;
         const chunks: ArrayBuffer[] = [];
 
@@ -128,7 +118,7 @@ export default function StreamAudio({ index = 0 }) {
             timeoutPromise(maxPromiseTime, promiseTimeoutMessage),
           ])) as ReadableStreamReadResult<ArrayBuffer>;
 
-          if (cacheTTS && value) {
+          if (shouldBufferChunks && value) {
             chunks.push(value);
           }
           if (value && mediaSource) {
@@ -137,39 +127,37 @@ export default function StreamAudio({ index = 0 }) {
           done = readerDone;
         }
 
+        mediaSource?.close();
+
         if (chunks.length) {
-          logger.log('Adding audio to cache');
-          const latestMessages = getMessages() ?? [];
-          const targetMessage = latestMessages.find(
-            (msg) => msg.messageId === latestMessage?.messageId,
-          );
-          cacheKey = targetMessage?.text ?? '';
-          if (!cacheKey) {
-            throw new Error('Cache key not found');
-          }
           const audioBlob = new Blob(chunks, { type });
-          const cachedResponse = new Response(audioBlob);
-          await cache.put(cacheKey, cachedResponse);
           if (!browserSupportsType) {
-            const unconsumedResponse = await cache.match(cacheKey);
-            if (!unconsumedResponse) {
-              throw new Error('Failed to fetch audio from cache');
-            }
-            const audioBlob = await unconsumedResponse.blob();
-            const blobUrl = URL.createObjectURL(audioBlob);
-            setGlobalAudioURL(blobUrl);
+            setGlobalAudioURL(URL.createObjectURL(audioBlob));
+          }
+          if (cacheTTS) {
+            const latestMessages = getMessages() ?? [];
+            const targetMessage = latestMessages.find(
+              (msg) => msg.messageId === latestMessage?.messageId,
+            );
+            cacheKey = targetMessage?.text ?? '';
+            if (!cacheKey) {
+              logger.warn('Cache key not found, skipping audio cache');
+            } else {
+              logger.log('Adding audio to cache');
+              await cache.put(cacheKey, new Response(audioBlob));
+            }
           }
-          setIsFetching(false);
         }
 
         logger.log('Audio stream reading ended');
       } catch (error) {
-        if (error?.['message'] !== promiseTimeoutMessage) {
+        if (error?.['message'] === promiseTimeoutMessage) {
           logger.log(promiseTimeoutMessage);
+          /** Let whatever was already appended finish playing instead of stalling forever */
+          mediaSource?.close();
           return;
         }
         logger.error('Error fetching audio:', error);
-        setIsFetching(false);
         setGlobalAudioURL(null);
       } finally {
         setIsFetching(false);
@@ -183,11 +171,10 @@ export default function StreamAudio({ index = 0 }) {
     setAudioRunId,
     setIsFetching,
     latestMessage,
-    isSubmitting,
     activeRunId,
     getMessages,
+    shouldPlay,
     isFetching,
-    audioRunId,
     cacheTTS,
     audioRef,
     voice,
diff --git a/client/src/components/Chat/Input/__tests__/AutoPlayAudio.spec.tsx b/client/src/components/Chat/Input/__tests__/AutoPlayAudio.spec.tsx
new file mode 100644
index 0000000000..252d9a28de
--- /dev/null
+++ b/client/src/components/Chat/Input/__tests__/AutoPlayAudio.spec.tsx
@@ -0,0 +1,56 @@
+import React from 'react';
+import { RecoilRoot } from 'recoil';
+import { render, screen } from '@testing-library/react';
+import AutoPlayAudio from '../AutoPlayAudio';
+import store from '~/store';
+
+jest.mock('../BrowserAudio', () => () => <div data-testid="browser-autoplay" />);
+jest.mock('../StreamAudio', () => () => <div data-testid="external-autoplay" />);
+
+const renderAutoPlayAudio = ({
+  engineTTS,
+  initialized = true,
+}: {
+  engineTTS: string;
+  initialized?: boolean;
+}) =>
+  render(
+    <RecoilRoot
+      initializeState={({ set }) => {
+        set(store.engineTTS, engineTTS);
+        set(store.speechSettingsInitialized, initialized);
+      }}
+    >
+      <AutoPlayAudio index={0} />
+    </RecoilRoot>,
+  );
+
+describe('AutoPlayAudio engine selection', () => {
+  it('drives autoplay through the Web Speech API for the browser engine', () => {
+    renderAutoPlayAudio({ engineTTS: 'browser' });
+
+    expect(screen.getByTestId('browser-autoplay')).toBeInTheDocument();
+    expect(screen.queryByTestId('external-autoplay')).not.toBeInTheDocument();
+  });
+
+  it('drives autoplay through the server stream for the external engine', () => {
+    renderAutoPlayAudio({ engineTTS: 'external' });
+
+    expect(screen.getByTestId('external-autoplay')).toBeInTheDocument();
+    expect(screen.queryByTestId('browser-autoplay')).not.toBeInTheDocument();
+  });
+
+  it('mounts nothing before speech settings initialize', () => {
+    renderAutoPlayAudio({ engineTTS: 'browser', initialized: false });
+
+    expect(screen.queryByTestId('browser-autoplay')).not.toBeInTheDocument();
+    expect(screen.queryByTestId('external-autoplay')).not.toBeInTheDocument();
+  });
+
+  it('mounts nothing for an unknown engine', () => {
+    renderAutoPlayAudio({ engineTTS: 'unsupported' });
+
+    expect(screen.queryByTestId('browser-autoplay')).not.toBeInTheDocument();
+    expect(screen.queryByTestId('external-autoplay')).not.toBeInTheDocument();
+  });
+});
diff --git a/client/src/components/Chat/Input/__tests__/BrowserAudio.spec.tsx b/client/src/components/Chat/Input/__tests__/BrowserAudio.spec.tsx
new file mode 100644
index 0000000000..3dcb631a07
--- /dev/null
+++ b/client/src/components/Chat/Input/__tests__/BrowserAudio.spec.tsx
@@ -0,0 +1,149 @@
+import React from 'react';
+import { RecoilRoot } from 'recoil';
+import { render, waitFor } from '@testing-library/react';
+import { Constants, QueryKeys } from 'librechat-data-provider';
+import { MemoryRouter, Routes, Route } from 'react-router-dom';
+import { QueryClient, QueryClientProvider } from '@tanstack/react-query';
+import type { TMessage, TConversation } from 'librechat-data-provider';
+import BrowserAudio from '../BrowserAudio';
+import store from '~/store';
+
+const conversationId = 'convo-1';
+const voiceName = 'Test Voice';
+const responseText = 'The capital of Germany is Berlin.';
+
+const spoken: string[] = [];
+let cancelCount = 0;
+
+class FakeSpeechSynthesisUtterance {
+  public voice: SpeechSynthesisVoice | null = null;
+  public onend: (() => void) | null = null;
+  public onerror: ((event: { error: string }) => void) | null = null;
+  constructor(public text: string) {}
+}
+
+const fakeVoice = { name: voiceName, localService: true } as SpeechSynthesisVoice;
+
+const speechSynthesis = {
+  getVoices: () => [fakeVoice],
+  addEventListener: () => undefined,
+  speak: (utterance: FakeSpeechSynthesisUtterance) => {
+    spoken.push(utterance.text);
+  },
+  cancel: () => {
+    cancelCount += 1;
+  },
+};
+
+const assistantMessage = {
+  messageId: 'assistant-1',
+  conversationId,
+  parentMessageId: Constants.NO_PARENT,
+  isCreatedByUser: false,
+  text: responseText,
+} as TMessage;
+
+const renderBrowserAudio = ({
+  activeRunId = 'run-1',
+  audioRunId = null,
+  isSubmitting = false,
+  messages = [assistantMessage],
+}: {
+  activeRunId?: string | null;
+  audioRunId?: string | null;
+  isSubmitting?: boolean;
+  messages?: TMessage[];
+} = {}) => {
+  const queryClient = new QueryClient({
+    defaultOptions: { queries: { retry: false } },
+  });
+  queryClient.setQueryData([QueryKeys.messages, conversationId], messages);
+
+  return render(
+    <QueryClientProvider client={queryClient}>
+      <MemoryRouter initialEntries={[`/c/${conversationId}`]}>
+        <RecoilRoot
+          initializeState={({ set }) => {
+            set(store.voice, voiceName);
+            set(store.conversationByIndex(0), { conversationId } as TConversation);
+            set(store.activeRunFamily(0), activeRunId);
+            set(store.audioRunFamily(0), audioRunId);
+            set(store.isSubmittingFamily(0), isSubmitting);
+          }}
+        >
+          <Routes>
+            <Route path="/c/:conversationId" element={<BrowserAudio index={0} />} />
+          </Routes>
+        </RecoilRoot>
+      </MemoryRouter>
+    </QueryClientProvider>,
+  );
+};
+
+/** Lets every effect settle so a missing utterance is a real absence, not a pending render */
+const settle = () => waitFor(() => expect(cancelCount).toBeGreaterThanOrEqual(0));
+
+describe('BrowserAudio autoplay', () => {
+  beforeAll(() => {
+    Object.defineProperty(window, 'speechSynthesis', {
+      writable: true,
+      configurable: true,
+      value: speechSynthesis,
+    });
+    Object.defineProperty(global, 'SpeechSynthesisUtterance', {
+      writable: true,
+      configurable: true,
+      value: FakeSpeechSynthesisUtterance,
+    });
+  });
+
+  beforeEach(() => {
+    spoken.length = 0;
+    cancelCount = 0;
+  });
+
+  it('speaks the finalized assistant message through speech synthesis', async () => {
+    renderBrowserAudio();
+
+    await waitFor(() => expect(spoken).toEqual([responseText]));
+  });
+
+  it('does not speak while the run is still submitting', async () => {
+    renderBrowserAudio({ isSubmitting: true });
+    await settle();
+
+    expect(spoken).toEqual([]);
+  });
+
+  it('does not speak a run that was already played', async () => {
+    renderBrowserAudio({ activeRunId: 'run-1', audioRunId: 'run-1' });
+    await settle();
+
+    expect(spoken).toEqual([]);
+  });
+
+  it('does not speak a message that is still streaming', async () => {
+    const streamingMessage = { ...assistantMessage, messageId: 'user-1_' } as TMessage;
+    renderBrowserAudio({ messages: [streamingMessage] });
+    await settle();
+
+    expect(spoken).toEqual([]);
+  });
+
+  it('does not speak the user message back to them', async () => {
+    const userMessage = { ...assistantMessage, isCreatedByUser: true } as TMessage;
+    renderBrowserAudio({ messages: [userMessage] });
+    await settle();
+
+    expect(spoken).toEqual([]);
+  });
+
+  it('cancels the utterance when the conversation is left', async () => {
+    const { unmount } = renderBrowserAudio();
+
+    await waitFor(() => expect(spoken).toEqual([responseText]));
+    unmount();
+
+    expect(cancelCount).toBeGreaterThan(0);
+  });
+});
diff --git a/client/src/hooks/Audio/MediaSourceAppender.ts b/client/src/hooks/Audio/MediaSourceAppender.ts
index 38c88ff188..cdcbf5467b 100644
--- a/client/src/hooks/Audio/MediaSourceAppender.ts
+++ b/client/src/hooks/Audio/MediaSourceAppender.ts
@@ -1,23 +1,56 @@
 export class MediaSourceAppender {
   private readonly mediaSource = new MediaSource();
   private readonly audioChunks: ArrayBuffer[] = [];
+  private readonly type: string;
 
+  private objectUrl?: string;
   private sourceBuffer?: SourceBuffer;
+  private isClosed = false;
 
   constructor(type: string) {
-    this.mediaSource.addEventListener('sourceopen', async () => {
-      this.sourceBuffer = this.mediaSource.addSourceBuffer(type);
+    this.type = type;
+    this.mediaSource.addEventListener('sourceopen', () => {
+      if (this.sourceBuffer != null) {
+        return;
+      }
 
+      this.sourceBuffer = this.mediaSource.addSourceBuffer(this.type);
       this.sourceBuffer.addEventListener('updateend', () => {
         this.tryAppendNextChunk();
       });
+
+      /** `sourceopen` only fires once a media element attaches the object URL, which
+       *  can land after the whole response was already read. Draining here is what
+       *  keeps short (fast) responses from stranding their chunks in the queue. */
+      this.tryAppendNextChunk();
     });
   }
 
   private tryAppendNextChunk() {
-    if (this.sourceBuffer != null && !this.sourceBuffer.updating && this.audioChunks.length > 0) {
-      this.sourceBuffer.appendBuffer(this.audioChunks.shift()!);
+    if (this.sourceBuffer == null || this.sourceBuffer.updating) {
+      return;
     }
+
+    const chunk = this.audioChunks.shift();
+    if (chunk == null) {
+      this.tryEndOfStream();
+      return;
+    }
+
+    this.sourceBuffer.appendBuffer(chunk);
+  }
+
+  private tryEndOfStream() {
+    /** `endOfStream()` only throws while the source is not open or a buffer is updating —
+     *  an empty response still has to end, or the element waits on it forever. */
+    if (!this.isClosed || this.mediaSource.readyState !== 'open') {
+      return;
+    }
+    if (this.audioChunks.length > 0 || this.sourceBuffer?.updating === true) {
+      return;
+    }
+
+    this.mediaSource.endOfStream();
   }
 
   public addBase64Data(base64Data: string) {
@@ -31,13 +64,17 @@ export class MediaSourceAppender {
     this.tryAppendNextChunk();
   }
 
+  /**
+   * Signals that no further data will arrive. The media source is only ended once every
+   * queued chunk has been appended, so the tail of the audio is never truncated.
+   */
   public close() {
-    if (this.mediaSource.readyState === 'open') {
-      this.mediaSource.endOfStream();
-    }
+    this.isClosed = true;
+    this.tryEndOfStream();
   }
 
   public get mediaSourceUrl() {
-    return URL.createObjectURL(this.mediaSource);
+    this.objectUrl ??= URL.createObjectURL(this.mediaSource);
+    return this.objectUrl;
   }
 }
diff --git a/client/src/hooks/Audio/__tests__/MediaSourceAppender.spec.ts b/client/src/hooks/Audio/__tests__/MediaSourceAppender.spec.ts
new file mode 100644
index 0000000000..635c0c2893
--- /dev/null
+++ b/client/src/hooks/Audio/__tests__/MediaSourceAppender.spec.ts
@@ -0,0 +1,164 @@
+import { MediaSourceAppender } from '../MediaSourceAppender';
+
+type Listener = () => void;
+
+class FakeSourceBuffer {
+  public updating = false;
+  public readonly appended: ArrayBuffer[] = [];
+  private readonly listeners = new Map<string, Listener[]>();
+
+  addEventListener(event: string, handler: Listener) {
+    const handlers = this.listeners.get(event) ?? [];
+    handlers.push(handler);
+    this.listeners.set(event, handlers);
+  }
+
+  appendBuffer(data: ArrayBuffer) {
+    this.updating = true;
+    this.appended.push(data);
+    queueMicrotask(() => {
+      this.updating = false;
+      this.listeners.get('updateend')?.forEach((handler) => handler());
+    });
+  }
+}
+
+const createdSources: FakeMediaSource[] = [];
+
+class FakeMediaSource {
+  public readyState: 'closed' | 'open' | 'ended' = 'closed';
+  public readonly sourceBuffers: FakeSourceBuffer[] = [];
+  private readonly listeners = new Map<string, Listener[]>();
+
+  constructor() {
+    createdSources.push(this);
+  }
+
+  addEventListener(event: string, handler: Listener) {
+    const handlers = this.listeners.get(event) ?? [];
+    handlers.push(handler);
+    this.listeners.set(event, handlers);
+  }
+
+  addSourceBuffer(_type: string) {
+    const sourceBuffer = new FakeSourceBuffer();
+    this.sourceBuffers.push(sourceBuffer);
+    return sourceBuffer;
+  }
+
+  endOfStream() {
+    this.readyState = 'ended';
+  }
+
+  /** A MediaSource stays `closed` until a media element attaches its object URL */
+  attach() {
+    this.readyState = 'open';
+    this.listeners.get('sourceopen')?.forEach((handler) => handler());
+  }
+}
+
+const flush = () => new Promise((resolve) => setTimeout(resolve, 0));
+const chunk = (byte: number) => new Uint8Array([byte]).buffer as ArrayBuffer;
+const appendedBytes = (source: FakeMediaSource) =>
+  (source.sourceBuffers[0]?.appended ?? []).map((buffer) => new Uint8Array(buffer)[0]);
+
+describe('MediaSourceAppender', () => {
+  let objectUrlCount = 0;
+
+  beforeEach(() => {
+    createdSources.length = 0;
+    objectUrlCount = 0;
+    Object.defineProperty(global, 'MediaSource', {
+      writable: true,
+      configurable: true,
+      value: FakeMediaSource,
+    });
+    Object.defineProperty(global.URL, 'createObjectURL', {
+      writable: true,
+      configurable: true,
+      value: () => `blob:media-source-${++objectUrlCount}`,
+    });
+  });
+
+  it('appends chunks that were queued before the media element attached', async () => {
+    const appender = new MediaSourceAppender('audio/mpeg');
+    const source = createdSources[0];
+    appender.mediaSourceUrl;
+
+    /** A short response can be read in full before `sourceopen` fires */
+    appender.addData(chunk(1));
+    appender.addData(chunk(2));
+    expect(appendedBytes(source)).toEqual([]);
+
+    source.attach();
+    await flush();
+
+    expect(appendedBytes(source)).toEqual([1, 2]);
+  });
+
+  it('appends chunks that arrive after the media element attached', async () => {
+    const appender = new MediaSourceAppender('audio/mpeg');
+    const source = createdSources[0];
+    appender.mediaSourceUrl;
+
+    source.attach();
+    appender.addData(chunk(1));
+    await flush();
+    appender.addData(chunk(2));
+    await flush();
+
+    expect(appendedBytes(source)).toEqual([1, 2]);
+  });
+
+  it('only ends the stream once every queued chunk has been appended', async () => {
+    const appender = new MediaSourceAppender('audio/mpeg');
+    const source = createdSources[0];
+    appender.mediaSourceUrl;
+
+    appender.addData(chunk(1));
+    appender.addData(chunk(2));
+    appender.close();
+    expect(source.readyState).toBe('closed');
+
+    source.attach();
+    await flush();
+
+    expect(appendedBytes(source)).toEqual([1, 2]);
+    expect(source.readyState).toBe('ended');
+  });
+
+  it('ends a closed stream that never received data', async () => {
+    const appender = new MediaSourceAppender('audio/mpeg');
+    const source = createdSources[0];
+    appender.mediaSourceUrl;
+
+    /** An empty response, or a read timeout before the first byte, still has to end —
+     *  otherwise the element waits on a source that can never receive data. */
+    source.attach();
+    appender.close();
+    await flush();
+
+    expect(source.readyState).toBe('ended');
+  });
+
+  it('ends a stream closed before the media element attached', async () => {
+    const appender = new MediaSourceAppender('audio/mpeg');
+    const source = createdSources[0];
+    appender.mediaSourceUrl;
+
+    appender.close();
+    expect(source.readyState).toBe('closed');
+
+    source.attach();
+    await flush();
+
+    expect(source.readyState).toBe('ended');
+  });
+
+  it('reuses a single object URL for the lifetime of the appender', () => {
+    const appender = new MediaSourceAppender('audio/mpeg');
+
+    expect(appender.mediaSourceUrl).toBe(appender.mediaSourceUrl);
+    expect(objectUrlCount).toBe(1);
+  });
+});
diff --git a/client/src/hooks/Audio/index.ts b/client/src/hooks/Audio/index.ts
index 15dfe6f746..fce85d220f 100644
--- a/client/src/hooks/Audio/index.ts
+++ b/client/src/hooks/Audio/index.ts
@@ -1,5 +1,6 @@
 export * from './MediaSourceAppender';
 export { default as useCustomAudioRef } from './useCustomAudioRef';
 export { default as usePauseGlobalAudio } from './usePauseGlobalAudio';
+export { default as useAutoplayTrigger } from './useAutoplayTrigger';
 export { default as useTTSExternal } from './useTTSExternal';
 export { default as useTTSBrowser } from './useTTSBrowser';
diff --git a/client/src/hooks/Audio/useAutoplayTrigger.ts b/client/src/hooks/Audio/useAutoplayTrigger.ts
new file mode 100644
index 0000000000..ec257808f8
--- /dev/null
+++ b/client/src/hooks/Audio/useAutoplayTrigger.ts
@@ -0,0 +1,37 @@
+import { useRecoilValue } from 'recoil';
+import type { TMessage } from 'librechat-data-provider';
+import { useLatestMessage } from '~/hooks/Messages/useLatestMessage';
+import { getLatestText } from '~/utils';
+import store from '~/store';
+
+export type TAutoplayTrigger = {
+  /** Whether the latest assistant message is finalized and its run has not been played yet */
+  shouldPlay: boolean;
+  activeRunId: string | null;
+  latestMessage: TMessage | null;
+};
+
+/**
+ * Shared "Autoplay Latest Message" gate so every TTS engine autoplays on identical terms:
+ * the run must be finished, the branch tail must be a persisted assistant message carrying
+ * text, and its run must not have been played already.
+ */
+export default function useAutoplayTrigger(index: string | number = 0): TAutoplayTrigger {
+  const activeRunId = useRecoilValue(store.activeRunFamily(index));
+  const audioRunId = useRecoilValue(store.audioRunFamily(index));
+  const isSubmitting = useRecoilValue(store.isSubmittingFamily(index));
+  const latestMessage = useLatestMessage(index);
+
+  const shouldPlay = !!(
+    !isSubmitting &&
+    latestMessage &&
+    latestMessage.isCreatedByUser !== true &&
+    getLatestText(latestMessage) &&
+    latestMessage.messageId &&
+    !latestMessage.messageId.includes('_') &&
+    activeRunId != null &&
+    activeRunId !== audioRunId
+  );
+
+  return { shouldPlay, activeRunId, latestMessage };
+}
diff --git a/client/src/hooks/Input/useTextToSpeechBrowser.ts b/client/src/hooks/Input/useTextToSpeechBrowser.ts
index b7361c5ff5..a88f126aeb 100644
--- a/client/src/hooks/Input/useTextToSpeechBrowser.ts
+++ b/client/src/hooks/Input/useTextToSpeechBrowser.ts
@@ -1,4 +1,4 @@
-import { useMemo, useSyncExternalStore } from 'react';
+import { useMemo, useCallback, useSyncExternalStore } from 'react';
 import { useRecoilValue } from 'recoil';
 import type { VoiceOption } from '~/common';
 import { subscribeSpeechVoices, getSpeechVoicesSnapshot } from '~/utils';
@@ -24,45 +24,52 @@ function useTextToSpeechBrowser({
     return filteredVoices.map((v): VoiceOption => ({ value: v.name, label: v.name }));
   }, [availableVoices, cloudBrowserVoices]);
 
-  const generateSpeechLocal = (text: string) => {
-    if (!isSpeechSynthesisSupported) {
-      console.warn('Speech synthesis is not supported');
-      return;
-    }
+  /** Reports whether an utterance was actually queued: autoplay must not mark a run as
+   *  played when the voice list has not loaded yet, or the message is never spoken. */
+  const generateSpeechLocal = useCallback(
+    (text: string): boolean => {
+      if (!isSpeechSynthesisSupported) {
+        console.warn('Speech synthesis is not supported');
+        return false;
+      }
 
-    const synth = window.speechSynthesis;
-    const voice = voices.find((v) => v.value === voiceName);
+      const synth = window.speechSynthesis;
+      const voice = voices.find((v) => v.value === voiceName);
 
-    if (!voice) {
-      console.warn('Selected voice not found');
-      return;
-    }
+      if (!voice) {
+        console.warn('Selected voice not found');
+        return false;
+      }
 
-    try {
-      synth.cancel();
-      const utterance = new SpeechSynthesisUtterance(text);
-      utterance.voice = synth.getVoices().find((v) => v.name === voice.value) || null;
-      utterance.onend = () => {
-        setIsSpeaking(false);
-      };
-      utterance.onerror = (event) => {
-        if (event.error === 'interrupted' || event.error === 'canceled') {
+      try {
+        synth.cancel();
+        const utterance = new SpeechSynthesisUtterance(text);
+        utterance.voice = synth.getVoices().find((v) => v.name === voice.value) || null;
+        utterance.onend = () => {
           setIsSpeaking(false);
-          return;
-        }
+        };
+        utterance.onerror = (event) => {
+          if (event.error === 'interrupted' || event.error === 'canceled') {
+            setIsSpeaking(false);
+            return;
+          }
 
-        console.error('Speech synthesis error:', event);
+          console.error('Speech synthesis error:', event);
+          setIsSpeaking(false);
+        };
+        setIsSpeaking(true);
+        synth.speak(utterance);
+        return true;
+      } catch (error) {
+        console.error('Error generating speech:', error);
         setIsSpeaking(false);
-      };
-      setIsSpeaking(true);
-      synth.speak(utterance);
-    } catch (error) {
-      console.error('Error generating speech:', error);
-      setIsSpeaking(false);
-    }
-  };
+        return false;
+      }
+    },
+    [isSpeechSynthesisSupported, voices, voiceName, setIsSpeaking],
+  );
 
-  const cancelSpeechLocal = () => {
+  const cancelSpeechLocal = useCallback(() => {
     if (!isSpeechSynthesisSupported) {
       return;
     }
@@ -74,7 +81,7 @@ function useTextToSpeechBrowser({
     } finally {
       setIsSpeaking(false);
     }
-  };
+  }, [isSpeechSynthesisSupported, setIsSpeaking]);
 
   return { generateSpeechLocal, cancelSpeechLocal, voices, isSpeechSynthesisSupported };
 }