ElevenLabs Voice Changer: Speech to Speech Explained

The ElevenLabs voice changer is a speech to speech tool: you record or upload your own performance and it re-voices it in a different voice while keeping your emotion, emphasis and delivery. It costs about 1,000 credits per minute of audio on subscription plans, or $0.12 per minute through the API, and works with library, cloned and designed voices.

Last updated: October 2026. Plans and credit costs checked against ElevenLabs’ official pricing page.

What is the ElevenLabs voice changer?

The ElevenLabs voice changer is a speech to speech feature that converts a recorded voice into a different target voice. Instead of typing a script, you supply the performance yourself, and ElevenLabs maps that delivery onto the chosen voice, so pauses, laughs and emphasis survive the conversion.

It sits alongside text to speech, dubbing and Voice Isolator in the ElevenLabs suite, described in the docs simply as a tool to modify and transform voices. The in-depth ElevenLabs review explains how it compares with the platform’s other audio tools.

“Modify and transform voices”

Source: ElevenLabs documentation

Put simply, the voice changer swaps the voice and keeps the acting.

How does speech to speech differ from text to speech?

Speech to speech differs from text to speech in its input: text to speech reads typed words and decides the delivery itself, while speech to speech takes a human performance and only changes the voice. That makes speech to speech better for nuanced acting and text to speech better for volume narration.

“A text-to-speech (TTS) system converts normal language text into speech”

Source: Wikipedia, Speech synthesis

Source: ElevenLabs pricing page and API pricing page (elevenlabs.io/pricing), checked October 2026.

Using the rough approximation of 1,000 characters per minute of speech, a minute of text to speech and a minute of voice changer cost about the same in credits, so the choice comes down to whether you want to perform the lines yourself. The ElevenLabs text to speech settings guide covers the typed route.

Pick speech to speech when the performance matters more than the typing speed.

What does the voice changer keep from your recording?

The voice changer keeps the performance layer of your recording and replaces only the vocal identity. According to ElevenLabs’ docs, it works without losing the nuances of the original, capturing whispers, laughs, cries, accents and subtle emotional cues that text to speech often struggles to reproduce from text alone.

Five elements of the source performance carry through to the new voice:

• Emotional delivery: Anger, warmth, sarcasm and hesitation recorded by the performer carry through to the target voice, which is the main reason actors prefer speech to speech.
• Whispers and laughs: Non-standard vocal sounds like whispers, laughs and cries are preserved, effects that are hard to coax out of a typed script reliably.
• Emphasis and pacing: Stressed words and the rhythm of a line follow the original performance, so comedic timing and dramatic pauses stay where the performer placed them.
• Accent cues: Subtle accent features in the recording are captured, though the target voice still shapes the final sound and may soften some regional traits.
• Optional noise reduction: A background noise removal setting reduces environmental sound in the output, helping recordings made outside a treated room sound cleaner after conversion.

Whatever the performer does, the new voice does too, so record the read you actually want.

How much does the ElevenLabs voice changer cost?

The ElevenLabs voice changer costs about 1,000 credits per minute of processed audio on subscription plans, the same rate as Voice Isolator. Monthly capacity therefore follows plan credits: roughly 10 minutes on Free, 30 on Starter, 121 on Creator and 600 on Pro if no credits go elsewhere.

Source: ElevenLabs pricing page (elevenlabs.io/pricing), checked October 2026. Minutes calculated by GrowthStackKit from ~1,000 credits per minute.

A 12-minute animated short voiced entirely through speech to speech would need about 12,000 credits, more than the Free plan’s monthly balance, while Creator could handle around ten such shorts a month.

Speech to speech is credit-hungry, so Creator is the realistic plan for regular character work.

Voice changer API pricing for developers

Voice changer API pricing is $0.12 per minute of audio, billed per usage rather than from subscription credits. That flat rate makes cost easy to forecast: a 10-minute file costs about $1.20 and an hour of converted dialogue costs about $7.20 before any other services are added.

The API references the target voice by its voice ID, so any cloned or designed voice in your account can be used programmatically. ElevenLabs’ docs recommend the multilingual speech to speech model over the English-only one, even for English content. The ElevenLabs API getting-started guide walks through keys and SDK setup.

For apps that convert user recordings, the API’s per-minute pricing is simpler to budget than credits.

How do you use the voice changer step by step?

Using the voice changer takes five steps in the ElevenLabs web app: pick the target voice, provide the source audio, adjust settings, generate and download. The whole process is quick, and most of the quality comes from the recording you feed in rather than from the settings.

1. Open Voice Changer: Selecting Voice Changer from the ElevenLabs dashboard opens a panel with a voice picker, an upload area and a record button for capturing audio directly.
2. Choose the target voice: Any voice in My Voices works, including library voices, instant or professional clones and voices created with Voice Design from a text prompt.
3. Record or upload audio: Recording a clean take in the browser or uploading an existing file supplies the performance, ideally with one speaker and minimal background noise.
4. Adjust settings: Enabling background noise removal and keeping the multilingual model selected suits most projects, while voice settings can fine-tune similarity to the target voice.
5. Generate and download: Generating the conversion, listening back and downloading the result completes the job, with retakes costing credits for each additional processed minute.

Long recordings must be split, because each conversion has a maximum segment length. A clean source take is the single biggest quality lever in this workflow.

Which target voice should you choose?

The target voice is the voice your performance will come out as, and ElevenLabs lets you pick from three sources. Library voices are instant, cloned voices reproduce a specific person with consent, and Voice Design creates a new synthetic voice from a written description of age, accent and tone.

For characters, browse the ElevenLabs voice library guide first, since saved library voices do not use custom voice slots. For a consistent brand narrator built from your own voice, see the comparison of instant and professional voice cloning. For fictional characters with no real-world match, ElevenLabs Voice Design fills the gap.

Match the target voice’s natural age and energy to your performance for the most convincing result.

How to record source audio that converts well

Source audio quality determines most of the voice changer’s output quality, because the model faithfully carries over everything in the recording, including noise and mumbling. A clean, close and confident take gives the target voice a clear performance to follow and avoids wasting credits on retakes.

Five recording habits make the biggest difference:

• Quiet recording space: A small furnished room with soft surfaces reduces echo and background hum that would otherwise be carried into the converted voice.
• Consistent microphone distance: Keeping the same distance from the microphone holds volume steady, so the target voice does not jump between loud and quiet passages.
• One speaker per file: Recording each character separately avoids crosstalk, letting each performance be converted to a different target voice with clean results.
• Committed performance: A confident, fully acted read converts better than a flat test take, because the target voice reproduces energy rather than adding any.
• Short segments: Breaking long scripts into scenes keeps each file under the per-conversion length cap and makes retakes cheaper when one line needs fixing.

Record like the final audio depends on it, because it does.

What are the best use cases for speech to speech?

Speech to speech works best where a human performance already exists or is easy to produce, and only the voice needs to change. Common uses include voicing multiple characters with one performer, keeping a character consistent across sessions and replacing specific words in an existing recording without re-recording the whole scene.

Five use cases come up most often:

• Multi-character animation: One performer voices an entire cast by acting each part and converting each line to a different target voice, keeping costs to a single session.
• Character consistency: A character keeps the same voice across episodes and languages even when recording conditions or the performer’s own voice change between sessions.
• Line fixes and pickups: Specific words or lines in an existing recording can be replaced by re-performing them and converting to the original voice for a seamless patch.
• Privacy for creators: Faceless creators can perform scripts naturally while publishing in a different voice, keeping their real voice private without sounding robotic.
• Game and app prototyping: Developers can mock up dialogue for many characters quickly, testing timing and tone before booking professional voice actors for final recordings.

Faceless creators combine this with other tools; the faceless YouTube video guide shows the full stack. Any job where the acting is good but the voice is wrong is a fit.

Can you use voice changer output commercially?

Voice changer output can be used commercially on paid plans: the pricing page lists a commercial license from Starter, while the Terms limit free use to non-commercial purposes. The Terms also say you retain all rights in and to your Output, which covers converted audio made under an eligible plan.

The bigger risk is the target voice, not the license. Converting a performance into a real person’s cloned voice requires their right and consent, which ElevenLabs asks users to confirm when creating a clone. The explainer on whether AI voice cloning is legal covers consent rules in more depth.

Paid plan plus consent for any real voice equals safe commercial use.

What are the voice changer’s limits?

The voice changer’s main limits are segment length, credit cost and source quality. ElevenLabs caps how long a single conversion can be, so longer recordings must be split, and at about 1,000 credits per minute it consumes credits faster per minute than transcription does.

It also cannot invent a performance that is not in the recording: a flat read stays flat. Accent changes are limited too, since the docs note that a clone’s accent and tone cannot be changed after creation, so pick a target voice that already has the accent you need. For cleaning audio without changing the voice, the ElevenLabs Voice Isolator tutorial is the right tool, at the same credit rate.

Split long files, start with a strong take and choose the target voice carefully.

Bottom line on the ElevenLabs voice changer

The ElevenLabs voice changer is the right tool when the performance matters: it preserves emotion, whispers and timing while swapping in a library, cloned or designed voice. At about 1,000 credits per minute it is credit-hungry, so Creator suits regular use, and developers get a simple $0.12 per minute API rate.

Speech to speech is how solo creators get a full cast without hiring one.

Source: GrowthStackKit, ElevenLabs Review

Test it on the Free plan’s roughly 10 minutes, then size your plan to the minutes you actually convert.

Frequently asked questions

The 12 most-asked questions about the ElevenLabs voice changer.

What does the ElevenLabs voice changer do?

The voice changer, also called speech to speech, converts a recorded or uploaded voice into a different target voice while keeping the original emotion, emphasis, whispers and laughs. It suits acting, character voices and line fixes, where a human performance matters more than typing a script for text to speech.

How many credits does the voice changer use?

The voice changer uses about 1,000 credits per minute of processed audio on subscription plans. That gives roughly 10 minutes a month on the Free plan’s 10,000 credits, about 30 minutes on Starter, about 121 minutes on Creator and about 600 minutes on Pro.

How much does the voice changer API cost?

Through the API, the voice changer costs $0.12 per minute of audio. A 10-minute file therefore costs about $1.20 and an hour of converted dialogue about $7.20. The target voice is referenced by its voice ID, so any cloned or designed voice in the account can be used.

Is speech to speech better than text to speech?

Neither is better overall. Speech to speech is better when nuanced acting matters, because it keeps the performer’s delivery. Text to speech is better for long narration and bulk content, because typing is faster than performing. Per minute, the credit cost is roughly similar using an approximate 1,000 characters per minute.

Can I use any voice as the target voice?

You can use voices from My Voices, including saved Voice Library voices, instant or professional clones and voices created with Voice Design. Cloning a real person’s voice requires confirming you have their right and consent, so use only your own voice or voices you are authorized to use.

Does the voice changer keep my accent?

ElevenLabs’ docs say the voice changer captures accents along with whispers, laughs and emotional cues, but the target voice still shapes the final sound. For a specific accent, choose a target voice that already has it, since a clone’s accent and tone cannot be changed after it is created.

Can the voice changer remove background noise?

Yes, there is a background noise removal setting that reduces environmental sounds in the converted output. For heavily noisy recordings, running Voice Isolator first can help, and recording in a quiet room with consistent microphone distance remains the most reliable way to get clean conversions.

Is there a length limit for the voice changer?

Yes. ElevenLabs sets a maximum segment length for each conversion, so longer recordings need to be split into shorter scenes. Splitting also makes retakes cheaper, since re-converting one short segment costs far fewer credits than reprocessing a long file at about 1,000 credits per minute.

Can I use voice changer audio on YouTube?

Yes on a paid plan. The pricing page lists a commercial license from Starter at $6 a month, and the Terms say you retain rights in your output. Audio made on the Free plan is limited to non-commercial use, which excludes monetized YouTube videos and client projects.

Does the voice changer work in other languages?

Yes. ElevenLabs’ docs recommend the multilingual speech to speech model over the English-only model, even for English content, so non-English performances can be converted too. The target voice and the quality of the source recording affect how natural the result sounds in each language.

What is the difference between Voice Changer and Voice Isolator?

Voice Changer replaces the voice in a recording with a different target voice. Voice Isolator keeps the original voice and removes background noise, ambient sound and music. Both cost about 1,000 credits per minute on subscription plans and $0.12 per minute through the API.

Can I try the voice changer for free?

Yes. The Free plan’s 10,000 monthly credits cover roughly 10 minutes of voice changer audio, enough to test a few scenes with different target voices. Free use is limited to non-commercial purposes, so upgrade to Starter or higher before publishing converted audio in monetized work.