Audio

feega generates audio with ElevenLabs. Every operation runs in one node type, Generate audio, and lands as an audio asset you can wire into other nodes or place on a motion timeline.

Operation	Input	Output	Billed per
Text to speech	text	audio	character
Voice changer	audio or video	audio	second
Dubbing	video or audio	dubbed video and audio, or audio	second
Music	text	audio	second
Sound effects	text	audio	second
Voice isolation	audio or video	audio	second

Also: voices (library, design, clone, manage) and voice-over and music in motion videos.

The Generate audio node

1. Add a Generate audio node from the add bar.
2. Pick the Operation. The inputs, the options and the ports change with it.
3. Write the text in the node, or wire a text node in. For audio and video operations, wire a media node in.
4. Press Generate.

If something is missing the node says what: Write the text or connect a text node, Connect an audio or video node, Pick a target language, Duration out of range.

Wire colour for audio is orange. Results are MP3 (44.1 kHz, 128 kbps), except a dubbed video, which stays a video.

Cost

Audio spends credits: the provider rate in dollars, times 120 credits per dollar, rounded.

Operation	Rate	Example
Text to speech (eleven_multilingual_v2, eleven_v3)	$0.08 per 1,000 characters	1,000 characters is 10 credits
Text to speech (eleven_flash_v2_5, eleven_turbo_v2_5)	$0.04 per 1,000 characters	1,000 characters is 5 credits
Voice changer	$0.12 per minute	1 minute is 14 credits
Dubbing	$0.50 per minute	1 minute is 60 credits
Music	$0.15 per minute	30 seconds is 9 credits
Sound effects	$0.12 per minute	5 seconds is 1 credit
Voice isolation	$0.12 per minute	1 minute is 14 credits

Per-second operations are billed on the length of the result (dubbing: the length the provider reports). A failed run is not billed. Org spend caps and the model allow-list apply as for any generation, see Generate and credits.

Agents, MCP and CLI

• MCP run_node_generation with medium: "audio" and params.operation set to text_to_speech, voice_changer, dubbing, music, sound_effects or voice_isolation. It never creates a node: create it first (insert_row) and pass its version.
• MCP describe_node_types with type: audio lists audio_operations (inputs, default model, price) and the voices.
• Dubbing comes back queued, like a video: poll the node.
• The CLI has no dedicated audio command. Use the MCP server.

If the server has no ElevenLabs key, audio generation answers elevenlabs_not_configured.