> For the complete documentation index, see [llms.txt](https://docs.dreamerland.ai/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.dreamerland.ai/kumi/kumi-voice.md).

# Kumi voice

Kumi can read its replies out loud. Choose one of the preset voices or upload a clip of your own voice. You can also give each persona its own voice, so every chat with that persona speaks the same way.

> Voice replies are free up to a daily quota. After that, each voice reply costs **1 extra credit**.

## Turn on voice replies

1. Open the **Chat** tab (the chat bubble icon in the bottom bar).
2. Tap the **⚙️ gear** at the top right to open **Chat settings**.

![The gear is at the top right of every chat](https://content.gitbook.com/content/tbrc6W2n35oVxftMovDQ/blobs/35XflJpKcs8zqLMVaWF7/01-chat-welcome.png)

3. Turn on **Voice replies**.

![Chat settings → Voice replies](https://content.gitbook.com/content/tbrc6W2n35oVxftMovDQ/blobs/lGE30T2jIaDNPsGR2tuH/02-chat-settings.png)

Kumi now reads each new reply aloud. Tap the **▶ play button** under a reply to hear it again.

* **Voice** chooses who is speaking.
* **Language** chooses the language Kumi speaks in.

Voice and language changes take effect right away. You don't need to tap Save.

> Voice replies are set per chat, and new chats start with them off. Turn them on in each chat where you want Kumi to talk.

## Choose a voice

Tap **Voice** to open **Choose a voice**.

![Choose a voice](https://content.gitbook.com/content/tbrc6W2n35oVxftMovDQ/blobs/Il2C3Q7o4vmQqLpPBVj4/03-choose-voice.png)

* **Upload my own voice / My voice:** your own voice clip (see below).
* **Persona voice:** the default voice of this chat's persona, if it has one.
* **Preset voices:** ready-made voices. Use the **Language** and **Gender** filters to narrow the list, and tap **▶** to preview a voice before choosing it.

Tap a voice to select it.

## Upload your own voice

You need a short audio file of someone talking. It should be at least **5 seconds** long, and the app uses a **5–20 second** part of it.

1. In **Choose a voice**, tap **Upload my own voice**.
2. Tap the upload box and pick an **MP3, WAV, M4A or AAC** file.

![Choose an audio file](https://content.gitbook.com/content/tbrc6W2n35oVxftMovDQ/blobs/IZQx2ugkzJQd52X3sQUJ/04-upload.png)

3. Drag the two handles on the waveform to select **5–20 seconds of clear speech**. Tap **▶** to listen to your selection.
4. *(Optional)* Under **What is said in this recording?**, type the exact words spoken in your selection. This helps Kumi copy the voice more accurately.
5. Tap **Done**.

![Pick 5–20 seconds of clear speech, then tap Done](https://content.gitbook.com/content/tbrc6W2n35oVxftMovDQ/blobs/4JuJKEKV0Wv7GSnPhwbM/05-upload-trim.png)

Your clip is saved as **My voice** and selected straight away. You can use it in any chat, and as a persona's default voice.

> You can keep one **My voice** at a time. Tap **Replace** to upload a different clip; it overwrites the old one.

**Tips for a good voice clip**

* One person speaking, close to the microphone.
* No music, background noise or echo.
* Talk naturally, at a normal pace.
* If you type the transcript, make it match the selected part word for word.

## Give a persona its own voice

Set a **default voice** on a persona, and every new chat with that persona starts with that voice.

1. Open the **Chat** tab and start a new chat. The **≡** menu at the top left has **New Chat**.
2. On the welcome screen, tap **PERSONA** to open **Personas**.
3. Tap **⋯** next to a persona and choose **Edit**. To make a new persona, tap **+** at the top or **Create new persona** (Pro).

![Personas → ⋯ → Edit](https://content.gitbook.com/content/tbrc6W2n35oVxftMovDQ/blobs/ex8dLda78osIZ8t0dW8O/06-personas-edit.png)

4. Scroll down to **DEFAULT VOICE**:
   * **Voice:** choose a preset or **My voice**. Tap **✕** to remove it.
   * **Language:** choose the language this persona speaks.
5. Tap **Save**.

![DEFAULT VOICE in the persona editor](https://content.gitbook.com/content/tbrc6W2n35oVxftMovDQ/blobs/kdE8MrrCyCF4uKICqTYR/07-persona-default-voice.png)

> Saving a persona's voice also updates **all existing chats** with that persona. You still turn **Voice replies** on in each chat.

## Change the voice in one chat

Tap **⚙️ → Voice** and pick another voice. This only changes the current chat; the persona keeps its default voice.

> If you later save a new voice on the persona, it replaces the voice in all of that persona's chats, including this one.

## FAQ

**Kumi isn't talking.** Check that **Voice replies** is on in this chat. While the voice is being made, a small loading icon appears under the reply.

**I see "Not enough credit to generate voice".** You've used today's free voice replies and have no credit left. Kumi still replies in text.

**My file was rejected.** The clip must be at least 5 seconds long and in MP3, WAV, M4A or AAC format. On the web version, the file itself must be 5–20 seconds long and 3 MB or smaller, because it can't be trimmed there.

**Can I save more than one own voice?** No, there is one **My voice**. Uploading again replaces it. For other voices, use the presets.
