Audio

Firelent turns a short description into finished spoken audio: a voiceover for your website, an explainer, a guided meditation, a welcome message — or a podcast-style conversation between two hosts. Firelent writes the script, casts the voices and records it.

How it works#

What you can make#

	Voices	Good for
Narration	One	Voiceovers, explainers, product intros, stories, meditations, welcome messages
Conversation	Two	Podcast episodes, interviews, Q&A, debates

Firelent picks the voices to match your description — “a calm, warm narrator”, “an energetic host and a thoughtful guest” — and the script follows the language you write in. Audio is speech only: music, sound effects and background soundscapes aren't possible yet. If you ask for them, Firelent records the voice part and tells you.

Length and cost#

• Say how long. Without a length, you get about two to three minutes. The maximum is 15 minutes.
• Firelent's voices speak about 170 words per minute, so a one-minute voiceover is roughly 170 words.
• 1 credit per started minute, at least 3. A five-minute episode costs 5 credits. You're only charged once the script is written, and you get your credits back if the recording fails.

Write a prompt that sounds good#

Create a 2-minute podcast conversation between two hosts about how small cafés can win more regulars: one host asks, the other shares three practical tips. Friendly and natural, with a short intro and outro.

• Give it a shape. For a podcast: a hook, the hosts, two or three segments, a short recap and a goodbye.
• Say how it should sound: calm, upbeat, serious, playful — and the pace, like “slow and relaxed” for a meditation.
• Ask for pauses where they help: “a short pause after each tip”. Firelent adds performance cues like pauses, laughs, warmth or a whisper where you ask for them.
• Give it the facts. Firelent can't look things up online or read files, so put names, dates and details in your message.
• Have your own text read: paste it and say “read it exactly as written”.

Write for the ear#

If you write the text yourself, write it to be heard, not read:

• Short sentences, one idea each — about 20 words or fewer.
• Round numbers: “over four thousand” is easier to follow than “4,180”.
• Signpost: “First…”, “The second tip…”, “To sum up…”.
• Spell out tricky names the way they're said: “Firelent — FIRE-lent”.
• Write numbers, dates and prices as they're spoken when it matters: “the third of May”, “twelve euros fifty”.

The player#

The player shows the title and the speakers, and lets you scrub through the recording.

Key	What it does
Space or K	Play and pause
J / L	Back / forward 10 seconds
← / →	Back / forward 5 seconds

Change it#

Ask for changes in the same chat — “make it slower”, “cut it to one minute”, “make the second host more sceptical”. Firelent writes and records a new version as a new file, at the usual price. It's written from your request and a summary of the previous version, so repeat anything that should stay the same. Earlier versions stay in the chat and in your Library.

You can't fix a single sentence on its own yet — each change records the whole piece again.

Use your audio#

• Download it from the chat or the player, or later from Library with the Audio filter. You get a standard audio file (MP3 or WAV) that plays everywhere.
• For a video: drop it into your video editor as the voiceover — see Videos.
• For a podcast: add your intro music and publish it with your podcast host.
• Tell people it's AI. If listeners could mistake the voices for real people, say that they're generated.

FAQ#

Can I pick a specific voice?#

Not by name. Describe the voice you want — age, energy, tone — and Firelent casts it.

Can I see the script?#

Not yet. If you need the text too, ask for a document with the same content.

Can I have more than two speakers?#

Not yet — one narrator or two speakers.

Next steps#