#Hi, I am using elevenlabs v3 to create a character and a few weeks ago I did text to speech and got

1 messages · Page 1 of 1 (latest)

steel ospreyBOT
#
AI-generated Answer for matttusonracing

ElevenLabs does not currently offer a feature to use a seed from previous generations to replicate or achieve similar results. Variations in output can occur even with the same settings due to the way the system processes audio generation.

To improve consistency and get results closer to what you liked, you can try the following:

  1. Adjust Stability Settings: Lower stability settings can make the voice more expressive but may introduce variability. Increasing stability might help maintain a consistent tone.

  2. Break Down Text: Divide your text into smaller segments, ideally under 1000 characters, to ensure better control over the voice output.

  3. Experiment with Style Settings: If you are using style exaggeration or other advanced settings, try adjusting them to see if they bring the output closer to your desired result.

  4. Use High-Quality Voices: If you are using a cloned voice, ensure the original sample is of high quality and the clone settings are optimized.

If these steps don't help, you might want to try regenerating the audio multiple times with slight adjustments to the text or settings to find a closer match.

sour orchid
#

I’ve tried every adjustment there is with V3 and I simply cannot get it remotely similar I used up all my credits trying to get something similar

steel ospreyBOT
#
AI-generated Answer for matttusonracing

It seems you're experiencing significant inconsistencies in audio outputs despite using the same settings, which can be frustrating. Unfortunately, there is no mention of a seed feature in the provided content that would allow you to replicate previous generations exactly. However, here are some suggestions to help you achieve more consistent results:

  1. Adjust Stability and Similarity Settings: Experiment with these sliders to refine the tone and flow of the voice. Stability can help maintain consistency in the voice's delivery.

  2. Break Text into Shorter Segments: Generating audio in smaller chunks can improve tone and emphasis, especially for complex or nuanced text.

  3. Ensure Proper Punctuation: Correct punctuation in your text can significantly impact the narration's tone and flow.

  4. Use Free Regeneration: If the prompt, voice, and model remain unchanged, and the first generation was made less than two hours ago, you can regenerate audio without using additional credits. Ensure the page hasn't been refreshed since the original generation.

If you've already tried these adjustments and are still facing issues, you may want to reach out to the support team for further assistance. They can investigate the problem and potentially offer solutions or compensation for the credits you've used. You can submit a ticket using the ticket form.

sour orchid
#

There is mention of a seed feature when I search online what is that talking about

steel ospreyBOT
#
AI-generated Answer for matttusonracing

Unfortunately, the provided content does not mention a seed feature or provide details about how to replicate specific generations in ElevenLabs v3. If you're struggling to recreate a voice generation, you might want to experiment further with text inputs, punctuation, or prompts to guide the delivery and style. For example, adding quotation marks around key phrases or using text that reflects the desired tone can sometimes help refine the output.

If you're still unable to achieve the desired result, I recommend reaching out to ElevenLabs' customer support team directly for assistance. You can submit a ticket using the Ticket Form. They may be able to provide more specific guidance or address your concerns regarding credits and generation consistency.

sour orchid
#

What effects do paragraphs have on delivery

steel ospreyBOT
#
AI-generated Answer for matttusonracing

Currently, ElevenLabs does not offer a seed feature for text-to-speech generations, so it is not possible to replicate a previous generation exactly using a seed. The AI determines the intonation and delivery based on the context, surrounding words, and punctuation in the text input. Proper punctuation is essential for achieving optimal results.

If you're struggling to achieve a similar output, you can try the following adjustments:

  1. Stability Settings: Lower stability can result in more dynamic speech patterns, while higher stability may lead to a more consistent but potentially monotonous output. Experiment with this setting to find a balance that works for your character.

  2. Style Exaggeration: Keep this setting at its default value to avoid issues like mispronunciations or inconsistent speech speed.

  3. Speech-to-Speech Feature: If available, this feature allows you to preserve tone and delivery by converting one voice to another.

  4. Projects Tool: For longer text, the Projects tool is recommended. It lets you generate content paragraph by paragraph, which can help maintain consistency and allows you to regenerate specific sections without reprocessing the entire text.

Regarding paragraphs, splitting text into shorter paragraphs can improve delivery and stability. Long sections of text may lead to inconsistent output, while shorter paragraphs allow for better control and easier corrections. Using the Projects tool also minimizes potential audio issues and lets you assign different speakers to specific paragraphs if needed.

If you're still unable to achieve the desired results, you may want to reach out to the customer support team for further assistance.

sour orchid
#

Where do I access the projects tool

steel ospreyBOT
#
AI-generated Answer for matttusonracing

To access the Projects tool, navigate to the Studio page on the ElevenLabs platform. This is where you can manage your projects, review settings, and work on long-form audio content. If you're on the free tier, note that the Projects feature is only available starting from the Starter plan.

sour orchid
#

Can I use V3 in projects I can’t seem to find an option

steel ospreyBOT
#
AI-generated Answer for matttusonracing

To access the Projects tool, you need to navigate to the 'Projects' page on the ElevenLabs platform. However, this feature is only available on paid plans starting from the Starter subscription. If you're currently on the free tier, you would need to upgrade your plan to regain access to Projects. You can find more details about subscription options on the Pricing Page.

Regarding using V3 in Projects, the content does not specify whether V3 voices are supported within the Projects tool. If you're unable to find the option, it may be worth confirming your subscription level or reaching out to the support team for clarification.

sour orchid
#

I am on the starter tier so I should have access to all

steel ospreyBOT
#
AI-generated Answer for matttusonracing

The Projects tool is not available on the Starter plan. To use this feature, you need to upgrade to the Creator plan or higher. You can find more details about subscription plans on the pricing page. Let me know if you have further questions!

sour orchid
#

You just said it was available from starter, but anyway can you use the V3 model in projects

steel ospreyBOT
#
AI-generated Answer for matttusonracing

The Projects feature is only available to users on the Creator tier or higher. If you're on the Starter tier, you won't have access to Projects, even though your existing projects will remain saved. To use this feature, you'll need to upgrade your subscription. You can check available plans on the Subscription Page.

Regarding the V3 model, it cannot be used in Projects if you don't have access to the Projects tool itself. If you believe there is an issue with your subscription tier or access, please contact the customer support team through the Ticket Form.

sour orchid
#

If I can generate paragraph by paragraph in projects how do I combine everything at the end

steel ospreyBOT
#
AI-generated Answer for matttusonracing

In the Projects module, you can generate audio paragraph by paragraph and then combine the audio into a single export. This allows you to create a cohesive output for your project. If you need to regenerate a specific paragraph, you can handle it manually before combining the audio.

sour orchid
#

Surely if I did regenerations paragraph by paragraph they’d all be different?

steel ospreyBOT
#
AI-generated Answer for matttusonracing

When generating audio paragraph by paragraph in the Projects tool, the output can vary slightly between sections due to differences in text structure, punctuation, and voice settings. However, you can minimize these differences by ensuring consistent settings for stability and similarity across all paragraphs. Adjusting these sliders to a balanced range (e.g., 40-60%) can help maintain a more uniform tone and delivery.

Once you've generated audio for each paragraph, you can combine them seamlessly within the Projects tool. The tool allows you to manage and export the entire project as a single audio file, ensuring smooth transitions between sections. Additionally, using the volume normalization feature during export can help maintain consistent audio levels across the combined file.

sour orchid
#

Also what voice tag can I use to adjust the pitch as the voice is too low pitched

steel ospreyBOT
#
AI-generated Answer for matttusonracing

Currently, ElevenLabs does not offer a built-in feature to adjust the pitch of a voice directly. To modify pitch, you would need to use external audio editing software during post-processing. Features for more control over audio elements like pitch are being considered for future updates, but no timeline is available.

For now, you can experiment with descriptive dialogue tags, such as "he said quietly" or "she spoke in a higher tone," to guide the AI's delivery style. This might help achieve subtle changes in the voice's tone or pitch.

sour orchid
#

How do you generate just one paragraph at a time its generated the lot when I try

steel ospreyBOT
#
AI-generated Answer for matttusonracing

To generate one paragraph at a time in the Projects tool, you need to highlight the specific paragraph you want to regenerate and then click the "Regenerate" button. This ensures only the selected paragraph is processed, rather than the entire text. Keep in mind that regenerating paragraphs will consume characters each time.

If you're combining paragraphs at the end, you can use the "Convert" button to create a new version of your project. This will save the full chapter or project, allowing you to download the completed audio file.

sour orchid
#

How do you change a project to using V3