#I'm using the elevellabs website and looking to use the "Voice Design" feature, but I can't seem to

1 messages · Page 1 of 1 (latest)

formal roverBOT
#
AI-generated Answer for solidaritea

Short answer: there isn’t a toggle. The Voice Design tool in the dashboard currently runs on Eleven v3 (Alpha). Voices you create with it are backward‑compatible and can be used with the v2 family of TTS models. (elevenlabs.io)

How to stay on v2 for production:

  • Create and save the voice with Voice Design (this uses v3 under the hood). (elevenlabs.io)
  • When you generate speech, select a v2 model in the TTS model dropdown (e.g., “Eleven Multilingual v2,” “Flash v2.5,” or “Turbo v2.5”). In the API, set model_id to eleven_multilingual_v2, eleven_flash_v2_5, or eleven_turbo_v2_5. (elevenlabs.io)

Why this matters: v3 is in public alpha and is not recommended for production; content from Beta/Alpha services cannot be used commercially. Using your designed voice with a v2 model keeps you on supported, production‑ready models. (help.elevenlabs.io)

Sources:
Voices | ElevenLabs Documentation
Voice design | ElevenLabs Documentation
Text to Speech (product guide) | ElevenLabs Documentation
What is Eleven v3 (Alpha)? – ElevenLabs

raven violet
#

Two follow up questions:

  1. Is there is no way for us to use the older Eleven V2 Voice Design Tool? I ask because if the voice design itself is V3 alpha and we are not supposed to rely on it because it may dramatically change.

  2. Is what you wrote here: "v3 is in public alpha and is not recommended for production; content from Beta/Alpha services cannot be used commercially. Using your designed voice with a v2 model keeps you on supported, production‑ready models" meant to imply that a Voice Designed in V3 when used on a V2 will NOT be at any risk of overhaul changes?

formal roverBOT
#
AI-generated Answer for solidaritea

Short answers up front

  1. On the website: there’s no documented toggle to use the old (v2) Voice Design in the dashboard. The dashboard Voice Design is described as powered by v3/Alpha. If you need v2 for designing, use the API and set model_id=eleven_multilingual_ttv_v2 on POST /v1/text-to-voice/design (this endpoint explicitly supports both eleven_multilingual_ttv_v2 and eleven_ttv_v3). (elevenlabs.io)

  2. A voice designed with Voice Design v3 is “backward compatible … across all our other models,” so you can use it with v2 models. However, v3 is in public alpha and “subject to change,” and Voice Design itself is marked “experimental.” The docs do not guarantee that a voice designed with v3 won’t change in future updates. Separately, content generated with Beta/Alpha services (including v3/Alpha) cannot be used commercially; content generated on paid plans with non‑Beta models (e.g., v2) can be used commercially. (help.elevenlabs.io)

Details and pointers

  • Dashboard Voice Design = v3/Alpha only (no UI switch documented). (elevenlabs.io)
  • API route to design with v2: POST /v1/text-to-voice/design with model_id=eleven_multilingual_ttv_v2 (allowed values: eleven_multilingual_ttv_v2, eleven_ttv_v3). (elevenlabs.io)
  • Backward compatibility of v3-designed voices with other models (including v2) is stated; stability is not guaranteed because v3 is Alpha/subject to change and Voice Design is experimental. (help.elevenlabs.io)
  • Commercial use: “content generated using v3 (Alpha)” cannot be used commercially; paid-plan generations not using Beta services can. (help.elevenlabs.io)

If you want, I can provide a minimal API example showing how to design with v2 and then synthesize with a v2 TTS model. (elevenlabs.io)

Sources:
Voices | ElevenLabs Documentation
What is Voice Design? – ElevenLabs
Voices | ElevenLabs Documentation
Design a voice | ElevenLabs Documentation
What is Voice Design? – ElevenLabs
What is Eleven v3 (Alpha)? – ElevenLabs
Design a voice | ElevenLabs Documentation

raven violet
#

Could you list the different features offered between the Website versus the API? I've never used the API before

formal roverBOT
#
AI-generated Answer for solidaritea

Here’s a concise, doc-sourced map of what’s available on the ElevenLabs website UI versus the API.

Website (no‑code tools and editors)

  • Studio: long‑form production UI with document/video import, timeline tracks (voices, captions, music, SFX), regenerate/lock paragraphs, collaboration and MP3/WAV export. Studio API exists but requires approval. (help.elevenlabs.io)
  • Dubbing Studio: create a Dubbing project, then interactively edit transcripts/translations and re‑render. UI accepts up to 500 MB / ~45 min per file. (help.elevenlabs.io)
  • Voice Library: browse and filter voices, add to My Voices, private/public sharing for Professional Voice Clones; search by uploading an audio sample to find the same/similar voices. (elevenlabs.io)
  • Voice Design UI: prompt to generate three options, then select/save one. (elevenlabs.io)
  • Voice cloning (UI): Instant Voice Cloning and Professional Voice Cloning wizards. (elevenlabs.io)
  • Voice Changer (speech‑to‑speech) UI: upload/record up to 5 minutes; costs 1,000 credits/min. (elevenlabs.io)
  • Sound Effects UI: generate four variants per prompt; up to 30s; looping; site pricing differs from API. (help.elevenlabs.io)
  • Voice Isolator UI: remove background noise; up to 500 MB / 1 hour; 1,000 credits/min. (help.elevenlabs.io)
  • Eleven Music UI: generate music with post‑generation section/lyrics editing. (help.elevenlabs.io)
  • AI Speech Classifier: free web tool to detect if audio was generated by ElevenLabs (notes current limitations). (elevenlabs.io)
  • Audio Native: embed an auto‑narration player on your site from the dashboard. (elevenlabs.io)
  • Agents platform (web): create/deploy conversational agents, connect tools/knowledge, manage integrations. (elevenlabs.io)

API and SDK (programmatic and realtime)

  • Text‑to‑Speech (HTTP): convert text with chosen voice/model; multiple output formats/bitrates. (elevenlabs.io)
  • TTS WebSockets (input streaming): send partial text and receive audio in realtime with alignment support. (elevenlabs-sdk.mintlify.app)
  • Voice Changer (speech‑to‑speech) API: transform source audio to a target voice; background‑noise removal and low‑latency PCM input options. (elevenlabs.io)
  • Speech‑to‑Text: Scribe v1 (batch/HTTP) and Scribe v2 Realtime (WebSocket) with commit/VAD strategies, timestamps/diarization options. (elevenlabs.io)
  • Dubbing API: create/list/get dubs, render outputs; API supports larger jobs (up to ~1 GB / ~2.5 hours).
raven violet
#

I'm not quite getting the rationale of warning people should not rely on anything V3 Alpha for long term projects, but then having the main website operate so that customers can only exclusively use the V3 Voice Design Tool; but I assume this isn't something an AI assistant can help me make any sense of lol