Text to Speech

Text to speech reads written text aloud and gives you an audio file. On wan2-6.net you paste the script, choose a voice and language, set the output format and say the line; the result can be downloaded and used wherever you need narration.

Write for the ear, not the page

Text that reads well silently often stumbles when spoken. A few habits fix most of it:

• Shorter sentences. A comma-heavy sentence that works on a page gives the voice nowhere to breathe.
• Spell out what should be said. Numbers, symbols and abbreviations are the usual source of a wrong reading: write "three point five million" if that is what you want to hear, not "3.5M".
• Punctuation is the director. Full stops and commas are what the voice uses for timing, so placing them deliberately is the cheapest way to improve the delivery.

Controls worth understanding

• Voice — a different voice changes the result more than any other setting. Try two before you tune anything else.
• Language — set it explicitly; a short line of text does not always make the language obvious.
• Speed, pitch and volume — small moves. Push the speed far above the default and the delivery flattens.
• Emotion — where the route supports it, this steers the tone; leave it unset if you are unsure, since a mismatched emotion is worse than a neutral read.
• Output format and sample rate — choose these for the destination. A high sample rate costs more and is wasted on a phone speaker.

Where it fits, and where it does not

It suits narration, interface prompts, placeholder audio for a cut, and reading a long document aloud. It is a poor fit for singing, for two voices in conversation, and for anything where the listener must hear a specific real person — that is a likeness question, not a synthesis one. Treat the first take as a test: check the pacing and any proper nouns before you build a longer script on top of it.

Audio can also be passed into a video route where the selected route accepts it, but synchronisation is not guaranteed — check the result rather than assuming the timing holds.

Questions people ask

Can I use the audio in a video? Where the route accepts an audio input you can pass it in; verify the sync yourself before publishing.

How much does it cost? The credit cost appears beside the form before you submit; longer scripts and higher sample rates cost more. Credit prices are on pricing.

Where do I start? Open the Wan workspace and choose the text-to-speech tab.