ElevenLabs Audio Tags: Make AI Voiceovers Sound Natural With Eleven v4

Learn ElevenLabs Audio Tags with Eleven v4.

  • Beginner
  • 6 steps
  • 1 resource
  • By The Rundown University
ElevenLabs Audio Tags: Make AI Voiceovers Sound Natural With Eleven v4

ElevenLabs released Eleven v4 on September 28, 2026. The model improves voice accuracy, delivery, emotion, and Audio Tags across more than 90 languages.

Audio Tags are short directions inside square brackets. They tell Eleven v4 how to deliver the words that follow. A tag can change emotion, volume, pace, or a human reaction. For example, [curious], [whispers], [slowly], and [sighs] can direct the same voice in different ways. ElevenLabs explains the available tag types and how to use them.

In this guide, you will:

  • Generate a plain short-form voiceover.
  • Add tags without changing the spoken words.
  • Compare the before and after audio.
  • Fix flat delivery or tags that get read aloud.
  • Create a short scene with two speakers.

The public samples were generated on October 15, 2026, using Eleven v4. The controlled single-speaker comparison uses the Roger voice with Stability at 50, Similarity at 75, and MP3 output at 44.1 kHz and 128 kbps. The live website returned two takes for each initial generation, and the players use Generation 1 from each run. ElevenLabs outputs can vary between generations.

Resources 1

Start from the exact file used in this guide instead of rebuilding it from scratch.

Unlock with Pro

$29 a month, cancel anytime

Step by step 6 steps

  1. Step 1 of 6

    Pick one voice and make the plain version

    Open Text to Speech in ElevenLabs.

    • Paste the plain script below.
    • Choose the Roger voice.
    • Select Eleven v4.
    • Set Stability to 50 and Similarity to 75.
    • Choose MP3, 44.1 kHz, 128 kbps.
    • Generate the audio and select Generation 1 from the two takes returned by the initial run.

    Use this script:

    Ever wonder why some AI voiceovers still sound flat? Every line gets the same delivery. The hook, the explanation, and the payoff all land at the same level. Eleven v4 audio tags let you direct emotion, pace, volume, and reactions inside the script. Start with one clear cue per clause. Keep the words natural. Then change the tag before you rewrite the line.

    ElevenLabs adds the generation to History automatically. Download the selected take, then keep the voice and settings unchanged for the tagged version.

    Eleven v4 with Roger selected and the plain script ready to compare.
    Eleven v4 with Roger selected and the plain script ready to compare.

  2. Step 2 of 6

    Add tags without changing the spoken words

    Paste this version. It uses the same spoken words. Only the bracketed directions differ.

    [curious, conversational] Ever wonder why some AI voiceovers still sound flat? [short pause] Every line gets the same delivery. [slightly faster, building energy] The hook, the explanation, and the payoff all land at the same level. [confident, clear] Eleven v4 audio tags let you direct emotion, pace, volume, and reactions inside the script. [thoughtful, slower] Start with one clear cue per clause. Keep the words natural. [warm, decisive] Then change the tag before you rewrite the line.

    Generate the tagged version with Roger, Eleven v4, Stability 50, Similarity 75, and MP3 at 44.1 kHz and 128 kbps. Select Generation 1 from the initial run.

    The same script with Audio Tags, Roger, and matching Eleven v4 settings.
    The same script with Audio Tags, Roger, and matching Eleven v4 settings.
    Listen Before: plain script
    Open audio file
    Read transcript

    Ever wonder why some AI voiceovers still sound flat? Every line gets the same delivery. The hook, the explanation, and the payoff all land at the same level. Eleven v4 audio tags let you direct emotion, pace, volume, and reactions inside the script. Start with one clear cue per clause. Keep the words natural. Then change the tag before you rewrite the line.

    Listen After: same script with Audio Tags
    Open audio file
    Read transcript

    Ever wonder why some AI voiceovers still sound flat? Every line gets the same delivery. The hook, the explanation, and the payoff all land at the same level. Eleven v4 audio tags let you direct emotion, pace, volume, and reactions inside the script. Start with one clear cue per clause. Keep the words natural. Then change the tag before you rewrite the line.

    What each tag asks for

    TagIntended job in the script
    [curious, conversational]Frames the opening as a direct question.
    [short pause]Gives the hook a moment to land.
    [slightly faster, building energy]Raises the pace through the list.
    [confident, clear]Directs a firmer delivery for the main point.
    [thoughtful, slower]Slows the practical advice.
    [warm, decisive]Directs a clear final instruction.

    Listen to both files without watching the script. Decide whether the hook, explanation, and final line are easier to distinguish. If the tagged delivery feels busy, remove one cue and generate another version.

  3. Step 3 of 6

    Fix a flat or overdone voiceover

    Use one change at a time. Keep each version in History so you can compare it with the last one.

    ProblemFirst fix to try
    The whole read stays flat.Use a voice that fits the job, then make one tag more specific. Try [curious, talking to one person] instead of [curious].
    The tag has little effect.Put it right before the clause it should change. Test a different voice if the delivery still misses.
    The read sounds too dramatic.Remove reaction tags, use plainer cues such as [calm] or [conversational], and raise Stability a little.
    The read sounds stiff.Lower Stability a little, keep the words natural, and generate another version.
    The same mood continues too long.Add a new tag at the exact point where the tone should change.
    The pace feels wrong.Try [slowly], [rushed], [short pause], or [long pause], then adjust punctuation.
    The voice changes too much between runs.Raise Stability a little and keep the same voice and Similarity setting.

    ElevenLabs notes that several generations may be needed. Keep the same test script while you compare tags or settings.

    Write tags that describe sound

    Clear directions reduce mix-ups. These pairs show how to make a vague cue more audible and specific.

    Vague cues

    [dark]
    [heroic]
    [sad scene]
    [fast]

    Clearer cues

    [low, restrained voice]
    [strong, proud delivery]
    [quiet, holding back tears]
    [rushed, short breaths]

    Eleven v4 can treat a vague cue as a delivery style or a sound effect. State the voice quality you want when the meaning could be unclear. The official prompting guide recommends clear audio direction.

  4. Step 4 of 6

    Fix Audio Tags being read aloud

    If the voice says the tag as part of the script, check these items in order:

    • Confirm Eleven v4 or Eleven v3 is selected.
    • Use square brackets: [excited].
    • Remove parentheses, braces, and quotation marks around the tag.
    • Put the tag next to the line it should affect.
    • Keep it as a short audio direction.
    • Generate the affected line or paragraph again.

    Wrong formatting

    (excited) This worked.
    {whisper} Keep this quiet.
    [standing near a window] I saw it.
    [gravelly] You came back.

    Better audio directions

    [excited] This worked.
    [whispers] Keep this quiet.
    [quiet, distracted] I saw it.
    [low, gravelly voice] You came back.

    ElevenLabs says tags may be read aloud when the model does not support them or the brackets are wrong.

    Pauses in Eleven v4

    Eleven v4 does not support SSML break tags. Use cues such as [short pause] or [long pause], plus commas and ellipses. If you need an exact pause length, add it in your audio or video editor after export. ElevenLabs documents the available pause methods.

  5. Step 5 of 6

    Create a two-speaker scene

    Eleven v4 supports multi-speaker dialogue. In the current Text to Speech website, use Add speaker to create six turns. Alternate Roger for Speaker 1 and Jessica for Speaker 2, then place each tag inside its turn. The website reuses the two selected voices as you add turns. ElevenLabs explains multi-speaker text-to-dialogue generation.

    Use this short scene in order.

    Speaker 1

    [skeptical, casual] You made this whole voiceover with AI?

    Speaker 2

    [matter-of-fact] The first version, yes.

    Speaker 1

    [teasing] And this version?

    Speaker 2

    [quietly, then more confident] I stopped asking for natural and started giving each line a clear direction.

    Speaker 1

    [curious] So the voice stayed the same?

    Speaker 2

    [warm] Same voice. Better performance.

    Generate the complete dialogue, then check:

    • Each line uses the intended speaker.
    • The voices remain distinct.
    • The question and reply use different delivery.
    • No tag is spoken aloud.
    • No extra words appear.
    • The pause between turns feels natural.
    Roger and Jessica assigned to alternating tagged dialogue turns in Eleven v4.
    Roger and Jessica assigned to alternating tagged dialogue turns in Eleven v4.
    Listen Two-speaker Eleven v4 dialogue
    Open audio file
    Read transcript

    You made this whole voiceover with AI? The first version, yes. And this version? I stopped asking for natural and started giving each line a clear direction. So the voice stayed the same? Same voice. Better performance.

    If one turn misses, revise its tag or voice assignment and generate the complete dialogue again. Keep approved wording and voice assignments fixed while you troubleshoot.

  6. Step 6 of 6

    Export and prepare the voiceover for video

    Download the selected clips from History. ElevenLabs records website generations there automatically. The Text to Speech product guide explains History and the available downloads. MP3 and WAV are available, with M4A and FLAC under advanced options.

    Use clear names:

    01-before-flat-v4.mp3
    02-after-tags-v4.mp3
    03-two-speaker-v4.mp3

    Keep a text file beside each clip with:

    • The full script.
    • Model and voice.
    • Stability and Similarity.
    • Output format.
    • Generation date.
    • The selected take.
    • Any change made after the first run.

    Before using the audio, listen for clipped words, unwanted sounds, wrong names, large volume jumps, and lines that do not match the script. Fix pronunciation with ElevenLabs' pronunciation tools or a clear phonetic spelling.

    The plain, tagged, and dialogue generations saved automatically in ElevenLabs History.
    The plain, tagged, and dialogue generations saved automatically in ElevenLabs History.

That's the whole workflow

Mark it complete to keep your progress. The background and common questions follow.

Keep learning with Rundown Pro

Guides stay free. This is what Rundown Pro adds:

  • Every course and recording
  • The Rundown Certified Operator program
  • Member perks

$29/mo, cancel anytime

Background and context

What are ElevenLabs Audio Tags?

Audio Tags are natural-language cues placed inside square brackets. Put a tag where the delivery should change:

[curious] Why does this voiceover still sound flat?

ElevenLabs does not limit you to a fixed list. You can combine cues, such as [whispering, nervous], or describe a clear speaking style, such as [calm, trying to sound reassuring]. The official Audio Tags guide includes more examples.

A few rules matter:

  • Put the tag before the words it should affect.
  • Add a new tag when the tone should change.
  • Keep each cue short and tied to something you can hear.
  • Use punctuation to help with rhythm.
  • Test the tag with the voice you plan to use.

The voice still matters more than the tag. Eleven v4 follows tags more reliably than earlier models, and a voice that already fits the job will usually give you a stronger starting point. ElevenLabs recommends matching the voice and prompt to the intended performance.

Eleven v4 vs Eleven v3

ElevenLabs recommends v4 for new work. V4 improves voice accuracy, consistency, emotion, delivery, and tag control. It also supports Professional Voice Clones and more than 90 languages. The Eleven v4 overview lists its current capabilities.

Scripts written with v3 tags should still work in v4. Test them again because v4 may deliver the same voice in a different way.

Eleven v4 uses Stability and Similarity controls. Lower Stability can allow more variation. Higher Stability keeps the voice closer to a fixed baseline. Higher Similarity holds closer to the reference voice, which can reduce some natural variation. V4 does not support the Style or Speed sliders, and it does not support SSML.

What you need

Use the ElevenLabs website and an account with Text to Speech access. You also need:

  • A voice you have the right to use.
  • Headphones or speakers.
  • A short script you can check word for word.
  • Enough credits for more than one generation.

The samples in this guide were generated on a verified paid Starter plan. ElevenLabs says commercial rights apply to content generated during a paid subscription, subject to the necessary rights, its terms, and applicable law.

Each generation can use credits. ElevenLabs allows up to two free website regenerations when the prompt or file, voice, and model stay the same, the first generation is less than two hours old, and you have not refreshed or left the page. Changing the tags changes the prompt, so the plain and tagged scripts count as separate generations.

For this test, skip Auto-tag. This exercise compares a plain script with directions you choose yourself.

ElevenLabs Audio Tags starter list

There is no complete list because Eleven v4 accepts natural-language cues. Use these copy-ready groups as starting points.

Emotion

[curious]
[excited]
[thoughtful]
[proud]
[nervous]
[playful]
[relieved]

Delivery

[conversational]
[confident]
[softly]
[quietly]
[whispers]
[low, restrained voice]

Pace

[slowly]
[rushed]
[short pause]
[long pause]
[building energy]
[drawn out]

Reactions

[laughs]
[chuckles]
[sighs]
[gasps]
[clears throat]
[exhales sharply]

Character

[sports commentator]
[calm teacher]
[tired detective]
[documentary narrator]

Sound events

[applause]
[footsteps]
[phone buzzing]
[door closing]
[rain]

For a normal voiceover, start with emotion, delivery, pace, and reactions. Add sound events only when you want them in the final track.

Frequently asked questions

Is there a full ElevenLabs Audio Tags list?

No. Eleven v4 accepts tags written in plain language, so the list is open-ended. Start with a known cue, then make it more specific when needed. ElevenLabs maintains a practical Audio Tags list and FAQ.

Can I make up my own tags?

Yes. Use a short cue that describes something the listener can hear. Try these examples:

[nervous, trying to sound confident]
[starting calm, then losing patience]

Do Eleven v3 tags work in Eleven v4?

Yes. ElevenLabs says v3 tags still work in v4, though the delivery may change. Test the script again with the same voice.

Do Audio Tags work through the API?

Yes. Add the tags inside the text and use eleven_v4, eleven_v4_turbo, or eleven_v3. Dialogue endpoints support separate voices for each turn. ElevenLabs documents model support in its Audio Tags FAQ.

Can I use SSML breaks with Eleven v4?

No. Use Audio Tags, punctuation, and text structure. Add exact silence in an editor when timing must be precise.

Why does the same tag sound different with another voice?

The voice's source audio, tone, and speaking style affect the result. Test the tag with the voice and language you plan to use. ElevenLabs' best-practices guide explains how voice choice affects delivery.