Text to Speech & Voice Changer
Text to Speech
Section titled “Text to Speech”Turns text into spoken audio — voice-overs, narration, scratch reads — using ElevenLabs voices.
| Inputs | text |
| Outputs | audio |
| Param | Notes |
|---|---|
model |
eleven_multilingual_v2 (default quality), eleven_turbo_v2_5 (fast), eleven_monolingual_v1. |
voice_id |
The ElevenLabs voice to speak with. |
stability |
Lower = more expressive variation between takes; higher = steadier read. |
similarity_boost |
How tightly the output clings to the voice’s timbre. |
style |
Style exaggeration — use sparingly; high values cost stability. |
speed |
Playback pace of the read. |
use_speaker_boost |
Extra similarity processing for the chosen speaker. |
Wire a Screenwriter or LLM output straight
into text and the script that was just written gets read aloud in the same
run.
Voice Changer
Section titled “Voice Changer”Speech-to-speech: keeps a performance — timing, emphasis, emotion — and replaces the voice. Record the read yourself, then deliver it in the campaign voice.
| Inputs | audio |
| Outputs | audio |
| Param | Notes |
|---|---|
model |
eleven_multilingual_sts_v2 or eleven_english_sts_v2. |
voice_id |
The target voice. |
stability / similarity_boost |
As in TTS — steadiness vs. timbre fidelity. |
remove_background_noise |
Clean the source recording before conversion. |
TTS or Voice Changer? If the read itself matters — comedic timing, a specific cadence — perform it and convert with Voice Changer. If you just need clean narration from text, TTS is one node shorter.