Generating Audio with AI in the Asset Manager

With the AI audio generator, you can create voice-overs directly in the asset manager. You choose a voice, give it a script, and get a finished audio file that's saved to your library, ready to use in your content.

You can type the script yourself, let the AI write a narration from a document, or dictate it with your microphone.

Before You Start

• Availability: AI audio generation is available if your organisation's licence includes it. It's then unlocked for everyone in your organisation. If you don't see Audio in the New menu, your licence doesn't include it yet. Contact us to upgrade your licence.
• Choose your folder first: Your finished audio is saved into the folder that's open in the asset manager. Open the target folder before you start.

Opening the Audio Generator

1. Open the asset manager and go to the folder where the audio should be saved.
2. Click New.
3. Choose Audio.

The Generate audio window opens. On the left you'll see the three steps you go through:

1. Voice: Listen and choose
2. Script: Type, upload or dictate
3. Result: Compare, tune, save

Use Continue and Back at the bottom to move between steps. Close at the top right closes the window at any time. Nothing is saved until you click Save to library.

Step 1: Choose a Voice

Every voice is shown as a card with:

• Play button: plays a short sample. Listen to a few voices before you decide.
• Name and badge: shows whether it's a Premade or Professional voice. Both come with the platform.
• Description and tags: for example gender, age, accent and language.

Click a card to select it, then click Continue.

Finding the right voice

• Search voices: searches voice names and descriptions.
• Source: All, Premade or Professional.
• Language, Gender, Age: these filters only list options that actual voices have.

If nothing matches, set the filters back to All or clear the search.

Tip: Choose a voice whose language matches your script. That usually sounds the most natural.

Step 2: Add Your Script

The voice you picked is shown at the top, so you can play its sample again. Below it, choose how you want to provide the script: Text, File or Speak.

Good to know: Text mode uses the least AI budget, because the AI only has to generate the voice. File and Speak mode use more, because the AI has to do extra work first. If you already have your script as text, paste it into Text.

Text: type or paste your script

Type or paste exactly what the voice should say, up to 5,000 characters. A counter below the field shows how many characters you've used.

When you're done, click Generate audio.

Emotion & delivery tags

In Text mode, you can add tags in square brackets to control how the voice speaks. For example:

[excited] Welcome to the course! [whispers] And here's a little secret…

Tags are highlighted in color in the text field. There are two ways to add them:

• Pick from the list: Put your cursor where the tag should go, then click a tag under Emotion & delivery tags. The tag is inserted right there.
• Type [ in the text: A suggestion list appears. Keep typing to narrow it down, then use the arrow keys and press Enter, or click a tag. Press Esc to close the list.

Tags are sorted into categories:

• Emotions: e.g. excited, nervous, calm, happy, curious
• Reactions: e.g. laughs, sighs, gasps, clears throat
• Delivery: e.g. whispers, shouts, cheerfully, sarcastic
• Pacing: e.g. pauses, hesitates, stammers
• Sound FX: e.g. applause, explosion. Sound effects are marked EXP (experimental). They don't always work, so treat them as a bonus.

Use Search all tags… to search every category at once.

You can also write your own directions, such as [speaks slowly]. The AI tries to follow them, but results can vary.

File: narrate a document

If your content already exists as a document, upload it. The AI reads it, writes a narration and speaks it.

Uses more AI budget: Before the voice is generated, the AI has to read and analyse your document and write the narration. That costs more than generating audio from text you type yourself. Each time you click Analyse & generate, the document is analysed again.

1. Click Upload a document and choose a PDF, Word or PowerPoint file.
2. Optional: tell the AI what you want under Instructions (optional), such as the focus, length or tone. For example: "Summarise this into a 60-second narration, friendly tone." If you leave it empty, the AI narrates the whole document.
3. Click Analyse & generate.

This takes longer than text. You'll see progress messages ("Analysing the document…", "Writing the narration…", "Generating the voice…"). Longer documents can take a minute or two, so keep the window open.

To swap the document, click the X next to the file name.

Speak: dictate your script

1. Click Record. The first time, your browser asks for permission to use your microphone. Allow it.
2. Read your script aloud. A timer shows how long you've been recording. Recording stops automatically after 3 minutes.
3. Click Stop & transcribe.

Your recording is turned into text and placed in the Text tab. There you can correct it and add tags before you click Generate audio.

Speak mode only turns your speech into text. The final audio is always spoken by the AI voice you chose, not by you.

Uses more AI budget: Your recording is first turned into text by the AI, and that transcription counts toward your AI budget on top of generating the voice. If you already have your script written down, Text is the cheaper choice.

Step 3: Compare, Fine-Tune and Save

This is where you listen to the result, try variations and save the version you want.

The selected take

At the top is a player for the take that's currently selected. Click the progress bar to jump to any point.

Below it:

• File name: the name the audio is saved under. It's pre-filled from the first words of your script or from your document's name. You can change it, and the file extension is added automatically.
• Transcript: shows the script that was spoken. For documents, this is where you can read the narration the AI wrote.

Takes: compare & select

Every generation and every effect adds a new take to the list. Each take shows:

• A play button, so you can compare takes quickly
• A label such as "Take 1 · Original" or "Take 2 · Volume +3 dB"
• The voice settings that were used

Select the take you want to keep with the radio button on the right. Only the selected take is saved. All other takes are discarded.

Voice settings

Open Voice settings to change how the voice performs:

• Stability (default 50%): Low sounds more expressive and varied. High sounds steadier and calmer.
• Similarity (default 75%): how closely the result sticks to the original voice.
• Style exaggeration (default 30%): strengthens the voice's typical style. Very high values can sound unnatural.
• Speaker boost (default on): makes the result sound more like the original speaker, but generating takes slightly longer.

Once you've changed a setting, click Regenerate voice. The whole script is spoken again from scratch and added as a new take. Effects already applied to the selected take are applied to the new take too.

The button is only active after you actually change a setting. Each regeneration uses a generation.

Audio effects

Under Audio effects, describe in your own words how the existing audio should change, and click Apply effect. For example:

• "Make it louder and a bit warmer"
• "Normalise the volume"
• "Slightly slower"

The effect is applied to the selected take and saved as a new take. Effects stack: if you apply another effect to that new take, it keeps the earlier ones. Effects are free, because they only adjust the existing audio and don't generate the voice again.

Saving

1. Select the take you want to keep.
2. Check the File name.
3. Click Save to library.

The audio is saved to the folder you had open, and you'll see the message "Audio saved to the library." The window then closes. You can now use the file in your content like any other audio file.

Step-by-Step Guides

A quick voice-over for a slide

1. Open the target folder, then click New > Audio.
2. Filter by Language, play a few samples and select a voice. Click Continue.
3. In Text, paste your script. Add a [pauses] where a short break helps.
4. Click Generate audio.
5. Listen, adjust the File name if needed, and click Save to library.

Turning a PDF or slide deck into a short summary

1. Choose a voice and click Continue.
2. Switch to File and upload your document.
3. Under Instructions (optional), write something like: "Summarise the key points in about 60 seconds, friendly and motivating tone, speak directly to the learner."
4. Click Analyse & generate and wait until it's done.
5. Open Transcript to check what the AI wrote.
6. If you want to change the wording, copy the transcript, go Back, switch to Text, paste it, make your edits and generate again. This also saves AI budget, because the document doesn't have to be analysed again.

Comparing two voices for the same script

1. Generate your first take as usual.
2. Click Back twice to return to the voice list and pick a different voice.
3. Click Continue. Your script is still there.
4. Click Generate audio again.

Both takes are now in the takes list. The list doesn't show which voice was used, so make a note of which take number belongs to which voice.

Making the voice sound more lively, or calmer

Change one or two settings at a time, click Regenerate voice, and compare the new take with the old one.

Tip: For emotion at specific moments, tags like [excited] or [calm] are often more precise than changing the settings for the whole script.

Fixing the volume or sound after generating

If the voice sounds right but is too quiet, too loud or too harsh, don't regenerate. Use Audio effects instead, for example "louder and normalised" or "reduce the harsh high frequencies". That's free and keeps the exact performance you liked.

Troubleshooting

• There's no "Audio" option under New: AI audio generation isn't enabled for your account. Contact your administrator.
• "Could not load voices": Close the window and open it again. If it keeps happening, contact support.
• Recording doesn't start: Check that your browser has permission to use the microphone. Some browsers don't support recording at all. In that case, use Text.
• "Could not generate audio from the document": Go back and try a shorter document, or paste the text into Text instead.
• Save doesn't work: Make sure a folder is open in the asset manager before you open the audio generator.