PRO feature.
Auto-generate on publish covers the posts you write from now on. This covers the archive you already have.
Instead of narrating thousands of old posts up front and hoping the right ones get listened to, let your readers decide. Once a post has been read a few times, TTSWP queues its audio in the background. Posts nobody opens never cost you a credit.
For a large site, this is the cheapest way to add audio where it actually gets used.
How to turn it on
- Go to Text to Speech → Content.
- Find Generate on first visit (queued) in the Automatic Audio Generation card.
- Flip it on.
- Set Visits before generating and your Daily credit cap.
- Click Save changes.
How a post qualifies
A page view only counts towards the threshold when all of these are true:
- The post has no audio yet in the language being viewed
- The post type is enabled on the Content page
- The visitor is a real browser, not a crawler or a bot
- It is an ordinary page view, and not a feed, search page, preview or REST request
Crawlers are excluded on purpose. A search engine visits every URL on your site once, so without that rule a single crawl would push your entire archive over the threshold at the same time.
See Control what gets narrated to choose which post types take part.
Your two spending controls
Visits before generating (3 by default) is how many qualifying visits a post needs before its audio is queued. Raise it if you only want your genuinely popular posts narrated. The lowest value is 1, which queues a post on its first real visit.
Daily credit cap (20,000 by default) is the most this feature is allowed to spend in a day. Once the cap is reached nothing further is queued until midnight, and the Content page tells you so. Set it to 0 to remove the limit.
Both limits are checked before anything is queued, so nothing ever enters the queue that you have not already decided you can afford.
What your visitors see
Audio is never generated while somebody is waiting for the page, so your site stays exactly as fast as it was.
That does mean the visitor who pushes a post over the threshold sees no player, because the audio does not exist yet. It is created moments later in the background, and every visitor after that gets the player as normal.
If you would rather no reader ever met a post without audio, use bulk generation or auto-generate on publish instead.
Credits
Audio queued this way is charged exactly like any other audio, and is drawn from your monthly balance first, then your top-up balance, then your own ElevenLabs key if you have connected one. See Credits explained.
Only audio that this feature queued counts towards its daily cap. Anything you generate manually, in bulk, or on publish is billed as usual and does not eat into the cap.
If a generation fails
A post whose generation fails is not tried again for 24 hours, so one problem post cannot burn an attempt every time somebody opens it. Once the underlying issue is fixed, the post becomes eligible again.
On publish, or on first visit?
| On publish | On first visit | |
|---|---|---|
| Covers | Posts you publish from now on | The archive you already have |
| Triggered by | You clicking Publish | Readers reaching your threshold |
| Spends credits on | Every new post | Only posts people actually read |
| Player on the very first view | Yes | No, from the second reader onwards |
The two work well together. Auto-generate on publish keeps new content covered from day one, while auto-generate on first visit fills in the back catalogue as demand appears.
Turning it off
Flip the toggle off. Audio that has already been generated stays exactly where it is, and anything already in the queue still finishes.