feat(audio): cap voice recording duration

Add an optional admin-configured duration limit for voice input and automatically submit the captured recording when the limit is reached.

Refs #28715

Assisted-by: ChatGPT 5.2

backend/open_webui/config.py

@@ -1552,6 +1552,10 @@ AUDIO_STT_ENGINE = os.getenv('AUDIO_STT_ENGINE', '')
 
 AUDIO_STT_MODEL = os.getenv('AUDIO_STT_MODEL', '')
 
+AUDIO_STT_MAX_RECORDING_DURATION = (
+    int(os.getenv('AUDIO_STT_MAX_RECORDING_DURATION')) if os.getenv('AUDIO_STT_MAX_RECORDING_DURATION') else None
+)
+
 AUDIO_STT_SUPPORTED_CONTENT_TYPES = [
     content_type.strip()
     for content_type in os.getenv('AUDIO_STT_SUPPORTED_CONTENT_TYPES', '').split(',')
@@ -3045,6 +3049,7 @@ DEFAULT_CONFIG = {
     'audio.stt.openai.api_request_format': AUDIO_STT_OPENAI_API_REQUEST_FORMAT,
     'audio.stt.engine': AUDIO_STT_ENGINE,
     'audio.stt.model': AUDIO_STT_MODEL,
+    'audio.stt.max_recording_duration': AUDIO_STT_MAX_RECORDING_DURATION,
     'audio.stt.supported_content_types': AUDIO_STT_SUPPORTED_CONTENT_TYPES,
     'audio.stt.allowed_extensions': AUDIO_STT_ALLOWED_EXTENSIONS,
     'audio.stt.azure.api_key': AUDIO_STT_AZURE_API_KEY,

backend/open_webui/main.py

@@ -2261,6 +2261,7 @@ async def get_app_config(request: Request):
         'audio.tts.voice',
         'audio.tts.split_on',
         'audio.stt.engine',
+        'audio.stt.max_recording_duration',
         'rag.file.max_size',
         'rag.file.max_count',
         'file.image_compression_width',
@@ -2363,6 +2364,7 @@ async def get_app_config(request: Request):
                     },
                     'stt': {
                         'engine': config.get('audio.stt.engine'),
+                        'max_recording_duration': config.get('audio.stt.max_recording_duration'),
                     },
                 },
                 'file': {

backend/open_webui/routers/audio.py

@@ -97,6 +97,7 @@ STT_CONFIG_KEYS = {
     'OPENAI_API_REQUEST_FORMAT': 'audio.stt.openai.api_request_format',
     'ENGINE': 'audio.stt.engine',
     'MODEL': 'audio.stt.model',
+    'MAX_RECORDING_DURATION': 'audio.stt.max_recording_duration',
     'SUPPORTED_CONTENT_TYPES': 'audio.stt.supported_content_types',
     'ALLOWED_EXTENSIONS': 'audio.stt.allowed_extensions',
     'WHISPER_MODEL': 'audio.stt.whisper_model',
@@ -256,6 +257,7 @@ class STTConfigForm(BaseModel):
     OPENAI_API_REQUEST_FORMAT: str = 'multipart'
     ENGINE: str
     MODEL: str
+    MAX_RECORDING_DURATION: Optional[int] = None
     SUPPORTED_CONTENT_TYPES: list[str] = []
     ALLOWED_EXTENSIONS: list[str] = []
     WHISPER_MODEL: str

src/lib/components/admin/Settings/Audio.svelte

@@ -51,6 +51,7 @@
 	let STT_OPENAI_API_REQUEST_FORMAT = 'multipart';
 	let STT_ENGINE = '';
 	let STT_MODEL = '';
+	let STT_MAX_RECORDING_DURATION: number | '' = '';
 	let STT_SUPPORTED_CONTENT_TYPES = '';
 	let STT_WHISPER_MODEL = '';
 	let STT_AZURE_API_KEY = '';
@@ -165,6 +166,9 @@
 				OPENAI_API_REQUEST_FORMAT: STT_OPENAI_API_REQUEST_FORMAT,
 				ENGINE: STT_ENGINE,
 				MODEL: STT_MODEL,
+				MAX_RECORDING_DURATION: STT_MAX_RECORDING_DURATION
+					? Number(STT_MAX_RECORDING_DURATION)
+					: null,
 				SUPPORTED_CONTENT_TYPES: STT_SUPPORTED_CONTENT_TYPES.split(','),
 				WHISPER_MODEL: STT_WHISPER_MODEL,
 				DEEPGRAM_API_KEY: STT_DEEPGRAM_API_KEY,
@@ -219,6 +223,7 @@
 
 			STT_ENGINE = res.stt.ENGINE;
 			STT_MODEL = res.stt.MODEL;
+			STT_MAX_RECORDING_DURATION = res.stt.MAX_RECORDING_DURATION ?? '';
 			STT_SUPPORTED_CONTENT_TYPES = (res?.stt?.SUPPORTED_CONTENT_TYPES ?? []).join(',');
 			STT_WHISPER_MODEL = res.stt.WHISPER_MODEL;
 			STT_AZURE_API_KEY = res.stt.AZURE_API_KEY;
@@ -262,6 +267,22 @@
 				</SettingsSelect>
 			</AdminSettingRow>
 
+			<AdminSettingField
+				label={$i18n.t('Maximum Recording Duration (seconds)')}
+				description={$i18n.t(
+					'Automatically stop and submit voice recordings after this many seconds. Leave blank for no limit.'
+				)}
+			>
+				<input
+					class={inputClass}
+					type="number"
+					min="1"
+					step="1"
+					bind:value={STT_MAX_RECORDING_DURATION}
+					placeholder={$i18n.t('No limit')}
+				/>
+			</AdminSettingField>
+
 			{#if STT_ENGINE !== 'web'}
 				<AdminSettingField
 					label={$i18n.t('Supported MIME Types')}

src/lib/components/chat/MessageInput.svelte

@@ -1597,6 +1597,7 @@
 					<div class={recording ? '' : 'hidden'}>
 						<VoiceRecording
 							bind:recording
+							maxDurationSeconds={$config?.audio?.stt?.max_recording_duration ?? 0}
 							onCancel={async () => {
 								recording = false;
 

src/lib/components/chat/MessageInput/VoiceRecording.svelte

@@ -20,6 +20,7 @@
 	export let echoCancellation = true;
 	export let noiseSuppression = true;
 	export let autoGainControl = true;
+	export let maxDurationSeconds = 0;
 
 	export let className = ' p-2.5 w-full max-w-full';
 
@@ -37,6 +38,9 @@
 	const startDurationCounter = () => {
 		durationCounter = setInterval(() => {
 			durationSeconds++;
+			if (maxDurationSeconds > 0 && durationSeconds >= maxDurationSeconds && recording && !loading) {
+				void confirmRecording();
+			}
 		}, 1000);
 	};