scriptoriumadmin/open-webui
1
0
Fork 0
mirror of https://github.com/open-webui/open-webui.git synced 2026-10-15 04:18:18 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 10

refac

Browse source
This commit is contained in:
Timothy Jaeryang Baek 2026-10-14 17:49:31 +04:00
parent 2bf3fb01ba
commit 3ec0774503
6 changed files with 168 additions and 48 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

4
backend/open_webui/config.py
View file

@ -1650,6 +1650,8 @@ AUDIO_TTS_VOICE = os.getenv('AUDIO_TTS_VOICE', 'alloy')
REALTIME_TTS_PROMPT_TEMPLATE = os.getenv('REALTIME_TTS_PROMPT_TEMPLATE')
AUDIO_REALTIME_ENABLED = os.getenv('AUDIO_REALTIME_ENABLED', 'False').lower() == 'true'
AUDIO_REALTIME_PROVIDER = os.getenv('AUDIO_REALTIME_PROVIDER', 'openai')
AUDIO_REALTIME_AUTH_TYPE = os.getenv('AUDIO_REALTIME_AUTH_TYPE', 'api_key')
AUDIO_REALTIME_OPENAI_API_BASE_URL = os.getenv('AUDIO_REALTIME_OPENAI_API_BASE_URL', 'https://api.openai.com/v1')
AUDIO_REALTIME_OPENAI_API_KEY = os.getenv('AUDIO_REALTIME_OPENAI_API_KEY', '')
AUDIO_REALTIME_MODEL = os.getenv('AUDIO_REALTIME_MODEL', 'gpt-realtime-2.1-mini')
@ -3127,6 +3129,8 @@ DEFAULT_CONFIG = {
'audio.tts.openai.params': AUDIO_TTS_OPENAI_PARAMS,
'audio.tts.api_key': AUDIO_TTS_API_KEY,
'audio.realtime.enabled': AUDIO_REALTIME_ENABLED,
'audio.realtime.provider': AUDIO_REALTIME_PROVIDER,
'audio.realtime.auth_type': AUDIO_REALTIME_AUTH_TYPE,
'audio.realtime.openai.api_base_url': AUDIO_REALTIME_OPENAI_API_BASE_URL,
'audio.realtime.openai.api_key': AUDIO_REALTIME_OPENAI_API_KEY,
'audio.realtime.model': AUDIO_REALTIME_MODEL,

6
backend/open_webui/routers/audio/__init__.py
View file

@ -12,7 +12,7 @@ import uuid
import wave
from fnmatch import fnmatch
from pathlib import Path
from typing import Optional
from typing import Literal, Optional
from urllib.parse import urlencode, urlsplit, urlunsplit
import aiofiles
@ -101,6 +101,8 @@ TTS_CONFIG_KEYS = {
REALTIME_CONFIG_KEYS = {
'ENABLED': 'audio.realtime.enabled',
'PROVIDER': 'audio.realtime.provider',
'AUTH_TYPE': 'audio.realtime.auth_type',
'OPENAI_API_BASE_URL': 'audio.realtime.openai.api_base_url',
'OPENAI_API_KEY': 'audio.realtime.openai.api_key',
'MODEL': 'audio.realtime.model',
@ -293,6 +295,8 @@ class STTConfigForm(BaseModel):
class RealtimeConfigForm(BaseModel):
ENABLED: bool = False
PROVIDER: Literal['openai', 'azure'] = 'openai'
AUTH_TYPE: Literal['api_key', 'microsoft_entra_id'] = 'api_key'
OPENAI_API_BASE_URL: str = 'https://api.openai.com/v1'
OPENAI_API_KEY: str = ''
MODEL: str = Field(default='gpt-realtime-2.1-mini', min_length=1, max_length=200)

104
backend/open_webui/routers/audio/realtime.py
View file

@ -15,6 +15,7 @@ from open_webui.models.config import Config
from open_webui.models.models import Models, ModelVoice
from open_webui.utils.access_control import has_permission
from open_webui.utils.auth import get_verified_user_by_token
from open_webui.utils.headers import get_microsoft_entra_id_access_token
from open_webui.utils.json_codec import JSONCodec
from open_webui.utils.models import check_model_access, get_all_models
from open_webui.utils.session_pool import get_session
@ -300,6 +301,55 @@ class CallProtocol:
raise ValueError('Unsupported call command')
async def _get_realtime_connection(config: dict) -> tuple[str, dict[str, str]]:
provider = config.get('audio.realtime.provider', 'openai')
auth_type = config.get('audio.realtime.auth_type', 'api_key')
if provider not in {'openai', 'azure'} or auth_type not in {'api_key', 'microsoft_entra_id'}:
raise ValueError('Unsupported Realtime provider or authentication method')
if provider != 'azure' and auth_type == 'microsoft_entra_id':
raise ValueError('Microsoft Entra ID requires the Azure OpenAI Realtime provider')
try:
url = urlsplit(config.get('audio.realtime.openai.api_base_url') or '')
valid_url = (
url.scheme in {'http', 'https'}
and url.hostname
and not (url.username or url.password or url.query or url.fragment)
)
url.port # Validate a supplied port before opening the connection.
except ValueError:
valid_url = False
if not valid_url:
raise ValueError('Invalid Realtime provider URL')
model = config.get('audio.realtime.model')
if not model:
raise ValueError('Configure the Realtime model')
ws_url = urlunsplit(
(
'wss' if url.scheme == 'https' else 'ws',
url.netloc,
url.path.rstrip('/') + '/realtime',
urlencode({'model': model}),
'',
)
)
if auth_type == 'microsoft_entra_id':
try:
token = await asyncio.to_thread(get_microsoft_entra_id_access_token)
except Exception:
raise ValueError('Unable to obtain a Microsoft Entra ID token for Realtime calls') from None
if not token:
raise ValueError('Unable to obtain a Microsoft Entra ID token for Realtime calls')
headers = {'Authorization': f'Bearer {token}'}
else:
key = config.get('audio.realtime.openai.api_key')
if not key:
raise ValueError('Configure the Realtime API key')
headers = {'api-key': key} if provider == 'azure' else {'Authorization': f'Bearer {key}'}
return ws_url, headers
@router.websocket('/realtime')
async def realtime_call(ws: WebSocket):
await ws.accept()
@ -321,6 +371,8 @@ async def realtime_call(ws: WebSocket):
raise ValueError('Authentication expired or invalid')
config = await Config.get_many(
'audio.realtime.enabled',
'audio.realtime.provider',
'audio.realtime.auth_type',
'audio.realtime.openai.api_base_url',
'audio.realtime.openai.api_key',
'audio.realtime.model',
@ -360,38 +412,30 @@ async def realtime_call(ws: WebSocket):
override = ModelVoice.model_validate((model_info.meta.model_dump().get('voice') if model_info else None) or {})
voice_model = config.get('audio.realtime.model')
voice = override.voice or config.get('audio.realtime.voice')
key = config.get('audio.realtime.openai.api_key')
url = urlsplit(config.get('audio.realtime.openai.api_base_url') or '')
if (
url.scheme not in {'http', 'https'}
or not url.netloc
or url.username
or url.password
or url.query
or url.fragment
):
raise ValueError('Invalid Realtime provider URL')
if not key or not voice_model or not voice or not config.get('audio.realtime.transcription_model'):
raise ValueError('Configure the Realtime API key, model, voice, and transcription model')
ws_url = urlunsplit(
(
'wss' if url.scheme == 'https' else 'ws',
url.netloc,
url.path.rstrip('/') + '/realtime',
urlencode({'model': voice_model}),
'',
)
)
session = await get_session()
if not voice_model or not voice or not config.get('audio.realtime.transcription_model'):
raise ValueError('Configure the Realtime model, voice, and transcription model')
async with asyncio.timeout(30):
async with asyncio.timeout(15):
upstream = await session.ws_connect(
ws_url,
headers={'Authorization': f'Bearer {key}'},
ssl=AIOHTTP_CLIENT_SESSION_SSL,
heartbeat=20,
max_msg_size=MAX_EVENT_BYTES,
)
ws_url, headers = await _get_realtime_connection(config)
session = await get_session()
try:
upstream = await session.ws_connect(
ws_url,
headers=headers,
ssl=AIOHTTP_CLIENT_SESSION_SSL,
heartbeat=20,
max_msg_size=MAX_EVENT_BYTES,
)
except aiohttp.WSServerHandshakeError as exc:
if exc.status in {401, 403}:
raise ValueError(
'Realtime provider authentication rejected. Check the credentials and access.'
) from None
if exc.status == 404:
raise ValueError(
'Realtime endpoint not found. Check the API base URL and deployment name.'
) from None
raise
event = await upstream.receive_json()
if event.get('type') != 'session.created':
raise ValueError('Provider did not create a voice session')

14
src/lib/apis/audio/index.ts
View file

@ -27,10 +27,22 @@ export const getAudioConfig = async (token: string) => {
return res;
};
export type RealtimeConfig = {
ENABLED: boolean;
PROVIDER: 'openai' | 'azure';
AUTH_TYPE: 'api_key' | 'microsoft_entra_id';
OPENAI_API_BASE_URL: string;
OPENAI_API_KEY: string;
MODEL: string;
VOICE: string;
TRANSCRIPTION_MODEL: string;
REALTIME_CALL_PROMPT_TEMPLATE: string | null;
};
type AudioConfigForm = {
tts: Record<string, unknown>;
stt: Record<string, unknown>;
realtime?: Record<string, unknown>;
realtime?: Partial<RealtimeConfig>;
};
export const updateAudioConfig = async (token: string, payload: AudioConfigForm) => {

79
src/lib/components/admin/Settings/Audio.svelte
View file

@ -8,9 +8,10 @@
getAudioConfig,
updateAudioConfig,
getModels as _getModels,
getVoices as _getVoices
getVoices as _getVoices,
type RealtimeConfig
} from '$lib/apis/audio';
import { config, settings } from '$lib/stores';
import { config } from '$lib/stores';
import Spinner from '$lib/components/common/Spinner.svelte';
import SensitiveInput from '$lib/components/common/SensitiveInput.svelte';
@ -31,8 +32,10 @@
export let saveHandler: () => void;
let realtime = {
let realtime: RealtimeConfig & { REALTIME_CALL_PROMPT_TEMPLATE: string } = {
ENABLED: false,
PROVIDER: 'openai',
AUTH_TYPE: 'api_key',
OPENAI_API_BASE_URL: 'https://api.openai.com/v1',
OPENAI_API_KEY: '',
MODEL: 'gpt-realtime-2.1-mini',
@ -146,7 +149,7 @@
try {
openaiParams = TTS_OPENAI_PARAMS ? JSON.parse(TTS_OPENAI_PARAMS) : {};
TTS_OPENAI_PARAMS = JSON.stringify(openaiParams, null, 2);
} catch (e) {
} catch {
toast.error($i18n.t('Invalid JSON format for Parameters'));
return;
}
@ -154,7 +157,7 @@
const res = await updateAudioConfig(localStorage.token, {
realtime: {
...realtime,
REALTIME_CALL_PROMPT_TEMPLATE: realtime.REALTIME_CALL_PROMPT_TEMPLATE.trim() || null
REALTIME_CALL_PROMPT_TEMPLATE: (realtime.REALTIME_CALL_PROMPT_TEMPLATE ?? '').trim() || null
},
tts: {
OPENAI_API_BASE_URL: TTS_OPENAI_API_BASE_URL,
@ -283,21 +286,67 @@
</SettingsSelect>
</AdminSettingRow>
{#if realtime.ENABLED}
<div class="grid grid-cols-1 gap-2 sm:grid-cols-2">
<AdminSettingField label={$i18n.t('OpenAI API Base URL')}>
<input class={inputClass} bind:value={realtime.OPENAI_API_BASE_URL} />
</AdminSettingField>
<AdminSettingField label={$i18n.t('API Key')}>
<SensitiveInput
variant="settings"
placeholder={$i18n.t('API Key')}
bind:value={realtime.OPENAI_API_KEY}
<AdminSettingRow label={$i18n.t('Provider')}>
<SettingsSelect
bind:value={realtime.PROVIDER}
ariaLabel={$i18n.t('Provider')}
on:change={() => (realtime.AUTH_TYPE = 'api_key')}
>
<option value="openai">{$i18n.t('settings.audio.realtime.openai')}</option>
<option value="azure">{$i18n.t('Azure OpenAI')}</option>
</SettingsSelect>
</AdminSettingRow>
{#if realtime.PROVIDER === 'azure'}
<AdminSettingRow
label={$i18n.t('Authentication')}
description={realtime.AUTH_TYPE === 'microsoft_entra_id'
? $i18n.t('settings.audio.realtime.entraHelp')
: ''}
>
<SettingsSelect bind:value={realtime.AUTH_TYPE} ariaLabel={$i18n.t('Authentication')}>
<option value="api_key">{$i18n.t('API Key')}</option>
<option value="microsoft_entra_id">{$i18n.t('settings.audio.realtime.entra')}</option>
</SettingsSelect>
</AdminSettingRow>
{/if}
<div class="grid grid-cols-1 gap-2" class:sm:grid-cols-2={realtime.AUTH_TYPE === 'api_key'}>
<AdminSettingField
label={$i18n.t(
realtime.PROVIDER === 'azure' ? 'settings.audio.realtime.azureUrl' : 'API Base URL'
)}
description={realtime.PROVIDER === 'azure'
? $i18n.t('settings.audio.realtime.azureUrlHelp')
: ''}
forId="realtime-api-base-url"
>
<input
id="realtime-api-base-url"
class={inputClass}
bind:value={realtime.OPENAI_API_BASE_URL}
/>
</AdminSettingField>
{#if realtime.AUTH_TYPE === 'api_key'}
<AdminSettingField label={$i18n.t('API Key')}>
<SensitiveInput
variant="settings"
placeholder={$i18n.t('API Key')}
bind:value={realtime.OPENAI_API_KEY}
/>
</AdminSettingField>
{/if}
</div>
<div class="grid grid-cols-1 gap-2 sm:grid-cols-2">
<AdminSettingField label={$i18n.t('Voice Model')}>
<AdminSettingField
label={$i18n.t(
realtime.PROVIDER === 'azure' ? 'settings.audio.realtime.deployment' : 'Voice Model'
)}
description={realtime.PROVIDER === 'azure'
? $i18n.t('settings.audio.realtime.deploymentHelp')
: ''}
forId="realtime-model"
>
<input
id="realtime-model"
class={inputClass}
bind:value={realtime.MODEL}
placeholder="gpt-realtime-2.1-mini"

9
src/lib/i18n/locales/en-US/translation.json
View file

@ -3908,5 +3908,12 @@
"Public key": "",
"Enter the Ed25519 public key provided by the app.": "",
"Edit details manually": "",
"Enter details manually": ""
"Enter details manually": "",
"settings.audio.realtime.openai": "OpenAI / compatible",
"settings.audio.realtime.entra": "Microsoft Entra ID",
"settings.audio.realtime.azureUrl": "Azure OpenAI API Base URL",
"settings.audio.realtime.azureUrlHelp": "Use the full API base URL, for example https://<resource>.openai.azure.com/openai/v1.",
"settings.audio.realtime.entraHelp": "Uses DefaultAzureCredential on the backend server, not the signed-in browser user. Configure a managed identity or Azure credentials on the server.",
"settings.audio.realtime.deployment": "Realtime Deployment Name",
"settings.audio.realtime.deploymentHelp": "Enter your Azure deployment name, which can differ from the model name. Choose a voice and transcription model supported by that deployment."
}