# Voice Changer

> Replace the voice in a recording or a talking video with another voice, and keep the original emotion, pacing and timing. Speech-to-speech with ElevenLabs.

Source: https://nodaro.ai/docs/nodes/audio/voice-changer

The **Voice Changer** node replaces the voice in a recording, or in a whole talking video, with another voice. It converts speech to speech with ElevenLabs, so the original emotion, pacing and timing stay and only the voice changes. Wire audio in to get audio back, or wire a video in to get the same video back with the new voice.

- Found in: Audio › Voices
- Output: audio
- API type: `voice-changer`

## When to use it
- Re-voice a talking-head or lip-synced video in one step.
- Give a recording the voice of your brand or of a character.
- Hide a speaker's identity and keep the natural delivery.
- Turn a rough scratch recording into a polished voiceover.
- Use one narrator voice across recordings of different people.

For a conversation with several speakers who each need a new voice, use [Voice Changer Pro](https://nodaro.ai/docs/nodes/audio/voice-changer-pro).

## Quick start
### Add the node

Press Tab on the canvas and choose **Audio › Voices › Voice Changer**.

### Wire the recording

Wire an audio node into the **Audio** input, or a video node into the **Video** input.

### Choose the new voice

Open the settings panel. Under **Voice**, open the voice browser and choose the voice to apply.

### Run it

Click **Run** on the node. For audio, the new recording appears on the **Audio** output. For a video, the re-voiced video appears on the **Video** output, and its new voice track on the **Audio** output.

Workflow: A talking clip gets a new voice; the video goes on to captions, and the new voice track goes on to a transcript.

- Upload Video → Voice Changer (video)
- Voice Changer → Add Captions (video)
- Voice Changer → Transcribe (audio)

## Inputs and outputs
| Input | Accepts | What it does |
| --- | --- | --- |
| **Audio** | Audio nodes, such as Upload Audio, Text to Speech and Voice Extractor | The recording whose voice is replaced. The node returns audio. |
| **Video** | Video nodes, such as Upload Video, Generate Video and Lip Sync | The talking video to re-voice. The node returns the video with the new voice. |

When both inputs are wired, the video wins and the audio input is ignored.

| Output | What it carries |
| --- | --- |
| **Audio** | The re-voiced audio. Always produced. For a video, this is the new voice track. |
| **Video** | The re-voiced video. Only produced when a video is wired in. |

**The file decides the mode, not the input.** The node checks what the file really contains. An audio-only file on the **Video** input, such as a voice memo saved as `.m4a`, runs as audio and returns audio. A video on the **Audio** input has its sound extracted and is treated as audio.

## Settings
| Setting | What it does |
| --- | --- |
| **Voice** | The target voice. Opens the voice browser with premade, Voice Library and your own voices. |
| **Model** | **Multilingual v2** (the default) covers 29 languages. **English v2** is for English only. Both cost the same. |
| **Stability** | From 0 to 1. The default is 0.5. Lower values are more expressive, higher values more uniform. |
| **Similarity** | From 0 to 1. The default is 0.75. How closely the result matches the target voice's timbre. |
| **Style Exaggeration** | From 0 to 1. The default is 0. Carries more of the source's delivery into the new voice, but adds latency and can reduce stability. |
| **Speaker Boost** | Strengthens the likeness to the target voice, with slightly more latency. On by default. |
| **Seed** | A whole number from 0 to 4,294,967,295. The same seed, input and settings give the same result. Leave it empty for a random result. |
| **Remove Background Noise** | **Off** (the default) keeps the music and sound effects under the new voice. **On** removes them for a clean, voice-only result. |

## Models
| Model | Maker | Modes | Credits | Details |
| --- | --- | --- | --- | --- |
| [ElevenLabs Voice Changer](https://nodaro.ai/docs/models/audio/elevenlabs-voice-changer) | ElevenLabs | Voice changer | 40 | Speech-to-speech: convert one voice to another while preserving prosody. |

Keep **Multilingual v2** unless you want to compare results. ElevenLabs recommends it even for English recordings, where it often does better than English v2, and it is required for recordings in other languages.

## Re-voice a whole video

Wire any talking video into the **Video** input: a generated clip, an uploaded file or a lip-synced shot. The node then:

1. Extracts the sound track from the clip.
2. Converts the speech to the voice you chose.
3. Puts the new voice back on the original video, and returns the video and the new voice track.

This replaces a chain of four nodes: a video, [Extract Audio](https://nodaro.ai/docs/nodes/video/extract-audio), a voice change and [Merge Video & Audio](https://nodaro.ai/docs/nodes/video/merge-video-audio).

**The video needs a sound track.** Most text-to-video and image-to-video models make silent video. Only models that generate sound, such as VEO 3.1 and Kling 3.0, give the node speech to convert. Wire audio directly when your clip is silent.

**Keep or drop the music.** With **Remove Background Noise** off, music and sound effects in the clip stay under the new voice. Turn it on when you want the voice alone.

## Tips
- **Start from the defaults.** Stability 0.5 and Similarity 0.75 suit most recordings. Higher similarity matches the voice more closely, but can sound less natural.
- **Leave Style Exaggeration at 0.** Raise it only when the new voice should lean harder into the drama of the original.
- **Clean the source first.** Poor recordings carry their problems into the result. Wire noisy audio through [Voice Extractor](https://nodaro.ai/docs/nodes/audio/voice-extractor) first.
- **Prefer speech-heavy clips.** Loud music under the dialogue can bleed into the conversion. Turn on **Remove Background Noise** when you do not need the music.
- **Expect less nuance on big changes.** The further apart the source and target voices are, for example a deep male voice to a light female voice, the more artifacts can appear.

## Troubleshooting
**The run fails with "This video has no audio track to revoice".** The clip is silent. Use a clip with spoken audio, or wire a recording into the **Audio** input.

**The result sounds robotic.** Lower **Similarity** to give the model more freedom, or choose a target voice that is closer to the original speaker.

## From the API
`POST /v1/voice-changer` takes `voiceId` and either `audioUrl` or `videoUrl`, plus the optional `model`, `stability`, `similarityBoost`, `style`, `useSpeakerBoost`, `seed` and `removeBackgroundNoise`. The result's `sourceHasVideo` tells you whether the run used video mode or audio mode. The MCP tool is `voice_changer`. See [Voice and media](https://nodaro.ai/docs/developers/api/voice-and-media).

## Frequently asked questions

### Does Voice Changer keep the original emotion and timing?

Yes. It converts speech to speech, so the pacing, intonation and emotion of the recording stay, and only the voice changes. A good original performance gives a good result.

### Can I change the voice of a video directly?

Yes. Wire the video into the Video input. The node returns the same video with the new voice, and the new voice track on its Audio output. The video must have a sound track with speech.

### Which model should I choose?

Keep Multilingual v2, the default. ElevenLabs recommends it even for English recordings, and it is required for other languages. English v2 is also available at the same price.

### Can I keep the background music under the new voice?

Yes. Leave Remove Background Noise off, which is the default, and the music and sound effects stay under the new voice. Turn it on for a clean, voice-only result.

### What is the difference between Voice Changer and Voice Changer Pro?

Voice Changer applies one voice to the recording. Voice Changer Pro detects each speaker in a conversation and gives each speaker a voice of its own.