The choices at the top of the page
Arnold's voice
Automatic — English: voice 18, German: voice 14 (recommended)
Voice 12 — English, best (B V3) (b0062db7)
Voice 16 — German, best (E V3) (d4abb987)
Voice 9 — English (B V0) (759681f6)
Voice 10 — English (B V1) (0a2d2b1e)
Voice 11 — English (B V2) (1bd0c9cb)
Voice 17 — English (D V0) (fe9246a9)
Voice 18 — English (D V1) (ee36d487)
Voice 19 — English (D V2) (32196a25)
Voice 20 — English (D V3) (57196258)
Voice 13 — German (E V0) (db9c9c58)
Voice 14 — German (E V1) (a5c1b140)
Voice 15 — German (E V2) (417673f8)
Voice 1 (17c57552)
Voice 2 (7a21ce82)
Voice 3 (37edca02)
Voice 4 (73a3354b)
ElevenLabs voice
Portrait
Answer style
Conversational — short, direct answers
In depth — his full persona, longer answers
Neutral — a factual assistant about Schoenberg
How much he speaks
Speak everything
Speak first part (the rest as text)
No voice (silent)
Big picture while he speaks
Off — portrait beside the conversation
On — portrait fills the screen while he speaks
How to follow his answers
Video + voice + text — the moving portrait speaks
Voice + text — his voice with the still picture (lighter)
Text only — just the written answers
More settings
Film look
On — like an old photograph Off
Face width
100%
Reset
Face height
100%
Reset
Portrait video
Simli live video
Still portrait (no video)
No portrait
Moving portrait
Anam (default)
Simli
Still picture (no movement)
No portrait
Video model (Simli)
fasttalk (Simli's default)
artalk (alternative — compare lip movement)
German titles & terms in English answers
Cartesia pronunciation dictionary (server-side; test)
As written: German in German; if the English is given too, just the English
Titles in English, music terms in German
Everything by its English name (smoothest)
Phonetic spelling (Cartesia, no pause)
Respelled for an English voice (old way)
Test: send the text as written, one piece — the voice model handles the German itself
Pause at a paragraph end (Cartesia)
Just the full stop — nothing extra sent to the voice
Extra half-second pause tag
Keep one voice per answer (Cartesia)
On — a piece in his other voice is made again; if that fails, its tone is filtered to match
On, but without the tone filter
Off — use every piece as it comes
Model
eleven_multilingual_v2 (default for the ElevenLabs voice)
eleven_turbo_v2_5 (deprecated name for flash v2.5)
eleven_flash_v2_5 (fast — same model, new name)
eleven_v3 (most expressive, not real-time)
eleven_v3_conversational (expressive + real-time)
eleven_v4 (newest, Sept 2026 — tested: added a British accent)
eleven_v4_turbo (newest — fast; tested: added a British accent)
sonic-3.6 (Cartesia — default)
sonic-3.5 (Cartesia)
sonic-3 (Cartesia — the original Jan 2026 snapshot; falls back to 3.5)
sonic-2 (Cartesia — no emotion/speed controls)
Cartesia emotion (overrides the server's setting)
Auto — vary per sentence (can change his voice mid-answer)
Steady — one tone for the whole answer
neutral
happy
excited
enthusiastic
elated
euphoric
triumphant
amazed
surprised
flirtatious
curious
content
peaceful
serene
calm
grateful
affectionate
trust
sympathetic
anticipation
mysterious
angry
mad
outraged
frustrated
agitated
threatened
disgusted
contempt
envious
sarcastic
ironic
sad
dejected
melancholic
disappointed
hurt
guilty
bored
tired
rejected
nostalgic
wistful
apologetic
hesitant
insecure
confused
resigned
anxious
panicked
alarmed
scared
proud
confident
distant
skeptical
contemplative
determined
Cartesia speed (overrides the server's setting)
server default
0.80
0.85
0.90
0.95
1.00
1.05
1.10
ElevenLabs variability (higher = more expressive, less consistent)
voice's dashboard setting
low (steady)
medium
high (expressive)
very high (wild)
ElevenLabs speed
voice's dashboard setting
0.85
0.90
0.95
1.00
1.05
1.10
Microphone
System default
✓ Done — back to the conversation
