Chatty
Phase 1 · loopback latency rig ·
two-way call →
Speak in
English
Chinese
Spanish
German
French
Japanese
Korean
Auto-detect (experimental)
Hear in
Spanish
Chinese (Simplified)
Chinese (Traditional)
English
German
French
Japanese
Korean
Start
Idle
—
ms speech-end → first audio
avg
—
p50
—
min
—
max
—
n
0
start→audio avg
—
jitter buf
—
underruns
0
session
0:00
reconnects
0
Settings
Transport
Direct (publisher token)
Via server proxy
Direct (session ws_url)
Send rate to Palabra
16 kHz (resampled)
24 kHz (resampled)
Native (no resample)
Silence to end segment (s)
0.5
Jitter buffer (ms)
150
Local VAD threshold
0.02
Voice
Playground voice (671e3dd0…)
default_low
default_high
Translate partial transcriptions (voices partial guesses, then re-voices the final — causes doubled speech)
Play through <audio> element (iOS silent-switch workaround)
Request screen wake lock
Log
Copy
Send log
Clear