StreamTranslate — Live Dashboard

Streamer / client: Do not use a headset or headphone mic for Voice Translation — use a USB or desk mic (or Elgato Wave Microphone, not Game). Headset drivers often hit Windows DirectSound errors. Keep Mic and Output on Default unless you deliberately route English TTS through VB-Cable for OBS. Urdu: Voice translation ke liye headphone / headset wala mic use mat karo — USB ya desktop mic (ya Wave Microphone) use karo. Mic aur Output ko Default par rakho jab tak OBS ke liye VB-Cable set na karna ho. Agar purani setting atki ho to config.json me mic_device_index aur output_device_index ko null karke app dubara chalao.

Settings runtime

Kick Channel

Dashboard language

OpenAI API key

DeepL API key

ElevenLabs API key

Google Translate API key

Use Windows default mic & speaker only (recommended)

On (default): no device picking here — set Input and Output in Windows Settings → Sound. Uncheck only if you need manual device lists below + Save.

Manual devices (optional)

Mic Device

This list is read from your PC. Pick your mic and Save — that input is what StreamTranslate uses (or Windows default if you leave “Default”).

Output Device

This list is read from your PC. Pick where TTS should play and Save — or Default for Windows playback. VB-Cable only if you route into OBS.

Mic device name (optional)

If set, overrides Mic dropdown / index after restart. Substring match on Windows device name.

TTS output device name (optional)

Target VB-Cable **Input** for ElevenLabs → OBS. Overrides Output dropdown when non-empty.

Streamer monitor (headphones) (optional)

Plays the same TTS again to this output after the primary device (sequential). Use for local monitoring while OBS takes VB-Cable.

Auto-fix wrong Windows default device (advanced)

Off (recommended): only the Mic and Output you choose above are used — or plain Windows default when “Default” is selected. Turn on only if sound still goes to the wrong device.

Discord collab (1 PC)

Recommended setup: two VB-Cable pairs (Cable A + Cable B)

One cable cannot cleanly separate “Discord call audio into the app” and “English TTS out to Discord mic” at the same time. Install a second VB-Audio virtual cable (official VB-Cable A+B / donation) or use VoiceMeeter — then map indices below from the device list.

VB-Audio Cable — add a second pair if you only have one.

Cable A — partner → you: Discord → Voice → Output = CABLE A Input (playback). StreamTranslate → Loopback capture = CABLE A Output (recording index).
Cable B — you → partner: StreamTranslate → Uplink playback = CABLE B Input (playback). Discord → Input (mic) = CABLE B Output (recording).
Listen: Listen playback = your Headphones index (Spanish TTS only in your ears).
Restart voice: Save → Voice OFF → ON (starts loopback thread).

Enable Discord loopback path (partner EN → your ears as ES)

Loopback capture index

Uplink playback index (EN TTS)

Listen playback index (ES TTS)

Loopback silence RMS

Loopback device name

Uplink output name

Listen playback name

Audience / stream: English only (hide Spanish transcript on dashboard & OBS subtitles)

Does not remove your spoken Spanish from the room — it stops broadcasting the Spanish text line to overlays and the live feed. For viewer audio, use only the VB-Cable English track in OBS. Latency breakdown: open /api/diagnostics (JSON) in a browser.

TTS voice (ElevenLabs)

Click to load all voices from your ElevenLabs account (cloned + library) into the dropdown above.

Custom Voice ID (clone / any library voice)

Leave empty to use the preset above. If you paste a Voice ID here, it overrides the dropdown when you Save. Use ElevenLabs Instant Voice Clone (or any voice), copy the ID, Save, then Test TTS.

TTS Stability: 0.5

TTS Similarity: 0.75

Test TTS

Browser Voice any device

Speak into this device's mic — audio is sent to the server, translated, and the result plays back here. Works from any PC/phone (and on the cloud server), no local audio setup needed.

Direction

Idle — click to record, speak, then click again to translate.

OBS browser sources

Use combined or two separate sources (chat + subtitles). For viewer audio: add OBS → Audio Input → VB-Audio CABLE Output and mix it in (not only your headphones).

Combined

http://localhost:5000/overlay

Chat only

http://localhost:5000/overlay/chat

Subtitles only

http://localhost:5000/overlay/subtitles

Same URLs on your LAN (replace localhost)

Combined

http://172.31.36.10:5000/overlay

Chat only

http://172.31.36.10:5000/overlay/chat

Subtitles only

http://172.31.36.10:5000/overlay/subtitles

Power

Stops this server completely. Start again with run.bat.

Kick — post to chat OAuth

Client flow (no curl): connect once below, then send a test message or turn on kick_post_translated_to_chat in config.json for auto-posted translations.

Connect Kick account

Loading…

Message to Kick chat

Prepend chat prefix from config (kick_post_chat_prefix)

Live Transcript

🇪🇸 ES → 🇺🇸 EN

🎙️

Start Voice Translation to see live transcripts here

Audio Level

Visual only: waves pulse after each transcript line above — not a live mic meter. If transcripts stay empty, check mic volume / Wave Link (Windows Settings → Sound → Input).

Pipeline & usage

Transcriptions

Translations

TTS Calls

Chat Msgs

—

Avg Latency ms

Whisper sec

Trans. chars

TTS chars

$0.00

Est. session (USD)

$0.00

Est. today (USD)

Cost is approximate — edit cost_* fields in config.json to match your provider invoices. Today’s total is stored in usage_daily.json.