Audio Recorder

Enable audio recording on MessageComposer to let users record voice messages.

import { MessageComposer } from "stream-chat-react";

const Composer = () => <MessageComposer audioRecordingEnabled />;

Once enabled, the composer renders StartRecordingAudioButton. When recording starts, MessageComposerUI swaps the normal composer UI for AudioRecorder.

Best Practices

Enable recording only where voice messages make sense for the product.
Use asyncMessagesMultiSendEnabled if users should be able to combine text, attachments, and voice recordings before sending.
Customize recording UI through WithComponents, not dead Channel override props.
Keep permission-denied messaging short and actionable.
Test both immediate-send and stacked-send behavior when changing recording flows.

UI Customization

The default recording surfaces are provided through ComponentContext:

StartRecordingAudioButton
AudioRecorder
RecordingPermissionDeniedNotification

Use WithComponents to override them:

import { Channel, MessageComposer, WithComponents } from "stream-chat-react";

const CustomStartRecordingAudioButton = (props) => (
  <button {...props}>Record</button>
);

const CustomAudioRecorder = () => <div>Custom recorder UI</div>;
const CustomRecordingPermissionDeniedNotification = () => (
  <div>Allow microphone access in your browser settings.</div>
);

const App = () => (
  <WithComponents
    overrides={{
      AudioRecorder: CustomAudioRecorder,
      RecordingPermissionDeniedNotification:
        CustomRecordingPermissionDeniedNotification,
      StartRecordingAudioButton: CustomStartRecordingAudioButton,
    }}
  >
    <Channel>
      <MessageComposer audioRecordingEnabled />
    </Channel>
  </WithComponents>
);

The custom components above are app-owned examples.

Browser Permissions

If microphone permission is denied when the user tries to start recording, the SDK opens a callout anchored to the recording button and renders RecordingPermissionDeniedNotification.

The default notification component receives only:

Prop	Type
`permissionName`	`"microphone" \| "camera"`

The close behavior is managed by the SDK dialog layer. Custom notification components do not receive an onClose callback.

Custom Encoding

By default, recordings are encoded as audio/wav. To reduce size, you can use the MP3 encoder plugin based on lamejs:

Install @breezystack/lamejs.

npm install @breezystack/lamejs

yarn add @breezystack/lamejs

Pass the encoder through audioRecordingConfig.

import { useMemo } from "react";
import { MessageComposer } from "stream-chat-react";
import { encodeToMp3 } from "stream-chat-react/mp3-encoder";

const Composer = () => {
  const audioRecordingConfig = useMemo(
    () => ({ transcoderConfig: { encoder: encodeToMp3 } }),
    [],
  );

  return (
    <MessageComposer
      audioRecordingConfig={audioRecordingConfig}
      audioRecordingEnabled
    />
  );
};

Sending Behavior

The recording uploads when recording is completed.

Sending behavior is controlled by asyncMessagesMultiSendEnabled on MessageComposer:

`asyncMessagesMultiSendEnabled`	Behavior
`false`	uploads and sends the voice recording immediately
`true`	keeps the voice recording in the composer preview stack until the user submits the whole composition

When a recording stays in the composer, it is rendered in the dedicated VoiceRecordingPreviewSlot.

Recording Controller

Components consuming MessageComposerContext can read recorder state through recordingController:

import { useMessageComposerContext } from "stream-chat-react";

const RecorderStatus = () => {
  const {
    recordingController: {
      completeRecording,
      permissionState,
      recorder,
      recording,
      recordingState,
    },
  } = useMessageComposerContext();

  return (
    <pre>
      {JSON.stringify(
        {
          hasRecorder: !!recorder,
          permissionState,
          recordingState,
          hasRecording: !!recording,
        },
        null,
        2,
      )}
    </pre>
  );
};

The controller exposes:

Property	Description
`completeRecording`	stops recording and finalizes the voice attachment
`permissionState`	current browser permission state
`recorder`	media-recorder controller instance
`recording`	generated voice-recording attachment after stop
`recordingState`	current recording lifecycle state