#When adding a alias to the Pronunciation Dictionaries in the elevenlabs settings. will it also work

1 messages · Page 1 of 1 (latest)

dark gateBOT
#
AI-generated Answer for robertschonholz

Short answer: yes—but only if you include that pronunciation dictionary in your API request.

  • For Text-to-Speech, pass the dictionary via pronunciation_dictionary_locators (array of {pronunciation_dictionary_id, version_id}) in the request body. Otherwise the API will not apply your Studio/Settings dictionary automatically. (elevenlabs.io)
  • You can retrieve the dictionary’s id and latest version_id with List/Get pronunciation dictionaries, then use those values in your request. (elevenlabs.io)
  • If you’re generating via Studio projects through the API, attach dictionaries to the project using the Update pronunciation dictionaries endpoint. (elevenlabs.io)
  • Compatibility note: alias rules work broadly; phoneme rules only take effect on specific models (e.g., eleven_flash_v2, eleven_turbo_v2, eleven_monolingual_v1). (elevenlabs.io)

Relevant docs:

  • Create speech (TTS) — pronunciation_dictionary_locators parameter. (elevenlabs.io)
  • List/Get pronunciation dictionaries. (elevenlabs.io)
  • Update pronunciation dictionaries on a project (Studio API). (elevenlabs.io)
  • Pronunciation dictionary model support details. (elevenlabs.io)

Sources:
Create speech | ElevenLabs Documentation
List pronunciation dictionaries | ElevenLabs Documentation
Update pronunciation dictionaries — ElevenLabs Documentation
Using pronunciation dictionaries | ElevenLabs Documentation
Create speech | ElevenLabs Documentation
List pronunciation dictionaries | ElevenLabs Documentation
Update pronunciation dictionaries — ElevenLabs Documentation
Using pronunciation dictionaries | ElevenLabs Documentation

atomic pond
#

does alias work with v3?

dark gateBOT
#
AI-generated Answer for robertschonholz

Short answer: Yes—if you mean the Pronunciations Editor’s “alias” (word substitution) rules in Studio. Those rules apply project-wide and aren’t model-limited; only phoneme tags have model restrictions, so alias substitutions work when generating with Eleven v3 (Alpha). (help.elevenlabs.io)

If instead you mean an SSML <sub alias="…"> tag via the API: ElevenLabs’ API docs only document support for <break> and SSML phoneme tags (and list which models support phonemes); they do not document an SSML alias tag, so it isn’t supported. (help.elevenlabs.io)

Sources:
How do I use the Pronunciations Editor in Studio? – ElevenLabs
Do pauses and SSML phoneme tags work with the API? – ElevenLabs

atomic pond
#

how to set the speed vor v3 text to speech?

dark gateBOT
#
AI-generated Answer for robertschonholz

Short answer:

  • Use model_id = "eleven_v3" and set voice_settings.speed. Valid range is 0.7–1.2 (default 1.0). Lower slows speech; higher speeds it up. This control works on all voices and models, including v3. (elevenlabs.io)

How to set it:

  • REST (Create speech): POST /v1/text-to-speech/:voice_id with JSON body including model_id: "eleven_v3" and voice_settings: { "speed": <value> }. (elevenlabs.io)
  • WebSocket streaming: in the initial message, include voice_settings with a speed field (e.g., "voice_settings": { "speed": 1.1 }). (elevenlabs.io)

Notes:

  • speed is part of the voice settings object (you can retrieve/see it via Voices → Get settings/default; defaults to 1.0). (elevenlabs.io)
  • You can also adjust Speed in the UI (Playground/Studio) under Voice settings; same 0.7–1.2 range. (elevenlabs.io)
  • Extreme values near the limits may degrade quality. (elevenlabs.io)

Sources:
Eleven v3 (alpha) — The most expressive Text to Speech model
Create speech | ElevenLabs Documentation
WebSocket | ElevenLabs Documentation
Get voice settings | ElevenLabs Documentation
Text to Speech (product guide) | ElevenLabs Documentation
Text to Speech (product guide) | ElevenLabs Documentation

atomic pond
#

emotion tags are getting ignored in text-to-speech. im using v3

dark gateBOT
#
AI-generated Answer for robertschonholz

Short answer: v3 does honor emotion “audio tags,” but they’re easy to neutralize if the setup or prompt isn’t aligned with v3’s guidelines.

Checklist to fix ignored tags:

  • Confirm you’re actually using the v3 model in your API call (model_id must be eleven_v3). (elevenlabs.io)
  • Use inline audio tags in square brackets immediately before or after the words they should affect, e.g., [whispers] That was close. Don’t place them far from the line they modify. (elevenlabs.io)
  • Adjust Stability: choose Creative or Natural. “Robust” makes v3 less responsive to directional prompts (tags), so they can appear ignored. (elevenlabs.io)
  • Pick a voice that fits the direction you’re giving. Tag effectiveness is voice‑dependent; v3 won’t reliably make a whispery voice shout just because you add [shout]. (elevenlabs.io)
  • Prefer designed/IVC voices over Professional Voice Clones for v3 right now; PVCs are not fully optimized for v3. (elevenlabs.io)
  • Use sufficiently long prompts. The v3 guide notes very short inputs are more likely to behave inconsistently; aim for >250 characters if possible. (elevenlabs.io)
  • If you’re new to tags, stick to the documented patterns like [whispers], [excited], [sighs], etc., and iterate near the target phrase. (help.elevenlabs.io)

If you’ve done the above and tags are still ignored, please share your exact model_id, voice, stability settings, and a short sample line; I can compare it against the v3 prompting guide. (elevenlabs.io)

Sources:
Eleven v3 (alpha) — The most expressive Text to Speech model
Prompting Eleven v3 (alpha) | ElevenLabs Documentation
Prompting Eleven v3 (alpha) | ElevenLabs Documentation
Prompting Eleven v3 (alpha) | ElevenLabs Documentation
Prompting Eleven v3 (alpha) | ElevenLabs Documentation
Prompting Eleven v3 (alpha) | ElevenLabs Documentation
How do audio tags work with Eleven v3 (Alpha)? – ElevenLabs
Prompting Eleven v3 (alpha) | ElevenLabs Documentation

atomic pond
#

its ignored in the elevenlabs app and api. i use v3 in both

dark gateBOT
#
AI-generated Answer for robertschonholz

Under v3, inline “audio tags” do work in both the app and API—but only when used exactly as documented and with a responsive voice/model setup. Please check:

  • Model: Ensure the model is set to eleven_v3 (or a current eleven_v3_preview_YYYY_MM_DD). In API, pass model_id: "eleven_v3" to the Create speech endpoint. In the app, select Eleven v3 (Alpha). (help.elevenlabs.io)
  • Tag syntax: Tags must be inline in square brackets and describe audible delivery or reactions, e.g. [whispers], [sad], [laughs]. Place them immediately before/after the phrase you want to affect. Avoid non-auditory tags like [standing] or [music]. Use lowercase tags as shown in docs. (elevenlabs.io)
  • Voice choice: During the alpha, Professional Voice Clones (PVCs) aren’t fully optimized for v3 and can respond weakly to tags. Try an Instant Voice Clone or a designed/library voice to verify tags are honored. (elevenlabs.io)
  • Settings (responsiveness): v3’s Stability strongly affects how well tags take effect. Use Creative/Natural (i.e., lower stability) for maximum expressiveness; Robust reduces responsiveness to directional prompts. (elevenlabs.io)
  • Minimal API example (verified pattern): POST /v1/text-to-speech/:voice_id with text like "[whispers] This is a test. [laughs] We made it." and model_id "eleven_v3". If this works but your original script doesn’t, the issue is likely voice/settings/prompting rather than tags. (elevenlabs.io)
  • App check: In the TTS Playground with v3 selected, paste a short line that uses tags exactly as above to confirm behavior end-to-end. v3 is a research preview and may require more prompt engineering; reliability can vary. (help.elevenlabs.io)

If tags are still ignored after the above, reset Style Exaggeration to 0, re-test, and then contact support with your request_id and sample text so we can investigate. (elevenlabs.io)

Sources:
What is Eleven v3 (Alpha)? – ElevenLabs
Eleven v3 (alpha) — The most expressive Text to Speech model
Eleven v3: Most Expressive AI Text to Speech Model Launched | ElevenLabs
Prompting Eleven v3 (alpha) | ElevenLabs Documentation
Create speech | ElevenLabs Documentation
What is Eleven v3 (Alpha)? – ElevenLabs
Troubleshooting | ElevenLabs Documentation

misty river
#

How to controll speed in V3

dark gateBOT
#
AI-generated Answer for zoja

Short answer:

  • Use the Speed setting. Valid range is 0.7–1.2 (default 1.0). Values <1.0 slow speech; >1.0 speed it up. Extreme values can affect quality. This applies to all models, including Eleven v3. (elevenlabs.io)

Where to set it:

  • API: pass speed in voice settings (e.g., voice_settings.speed = 0.85) on your TTS request; you can also read defaults which include "speed". (elevenlabs.io)
  • Studio / Agents / Playground UI: adjust Speed under Voice settings. (elevenlabs.io)

V3-specific notes:

  • Eleven v3 also responds to pacing cues via audio tags (e.g., [rushed], [slows down]) and pauses ([pause], [breathes]); punctuation and breaks can further shape timing. Use these alongside the Speed setting for finer control. (elevenlabs.io)

Related docs:

Sources:
Speed control | ElevenLabs Documentation
Create speech | ElevenLabs Documentation
Studio overview | ElevenLabs Documentation
Prompting Eleven v3 (alpha) | ElevenLabs Documentation
Speed control | ElevenLabs Documentation
Can I change the pace of the voice? – ElevenLabs
Prompting Eleven v3 (alpha) | ElevenLabs Documentation
ElevenLabs v3 Audio Tags: Precision Delivery Control | ElevenLabs
Get voice settings | ElevenLabs Documentation

misty river
#

"Studio / Agents / Playground UI: adjust Speed under Voice settings." - gives me "page not found"