Generate ElevenLabs speech with timestamps
Approval-gated ElevenLabs workflow for generating one text-to-speech response as JSON with base64 audio and character alignment timestamps. It reads the selected voice and default settings first, then calls recued-core.elevenlabs.speech.create with explicit text, model, output format, and voice settings. It does not download binary audio, stream output, upload audio, clone voices, train PVC voices, mutate voices or samples, configure agents/tools/MCP/secrets, or call arbitrary ElevenLabs routes.
How it works
Inspect the data fetches, transforms, gates, and output this recipe runs.
Settings
Configurable at install. Defaults shown — change them anytime in Recued.
text
setting
=
speed
setting
=
1
style
setting
=
0
reason
setting
=
model id
setting
=
eleven_multilingual_v2
voice id
setting
=
stability
setting
=
0.5
elevenlabs
setting
=
output format
setting
=
mp3_44100_128
enable logging
setting
=
true
similarity boost
setting
=
0.75
use speaker boost
setting
=
true
apply text normalization
setting
=
auto
apply language text normalization
setting
=
false
Trust & control
What installing this recipe would let it do. Recued grants these permissions at install — review them there before approving.