The Logs page records the last 50 requests the plugin sent to OpenAI, so you can see what was generated, by whom, and whether each request succeeded. Go to AI Text-to-Speech › Logs. Only administrators can open it.

Columns in the log

• Date & Time: when the request was made, in your site’s date and time format.
• Post ID: links to the post, or shows “Admin” for Generator page requests.
• User: the username of the person who clicked generate.
• Voice Model, Voice and Voice Style.
• Total Characters: characters sent in that request.
• Estimated Cost: characters multiplied by the per-1,000 rate for the model currently selected in settings ($0.012 or $0.015). If you’ve changed model since, older rows are recalculated at the new rate.
• Response Code: 200 (Success) in green, 429 (Rate Limit) in orange, or another HTTP code.

Why one post has several rows

Each row is one API request, not one audio file. Long posts are split into chunks (up to 6,000 characters each on GPT-4o mini TTS, 2,000 on TTS-1), and the spoken intro and each [voice] section are sent separately. A long article can therefore create five or more rows. Add the Total Characters for all rows from the same generation to get the full size.

Because the log only holds 50 rows, a few long posts can push older entries out quickly. For a full billing history, use the usage dashboard in your OpenAI account.

Failed requests and the error log

Rows are written after a generation finishes. If a request fails part way, generation stops and the error is shown in an alert in the editor, but that attempt may not appear in this table.

Failures are recorded separately in the Error Log box on AI Text-to-Speech › Settings › General Settings. It holds the most recent 10 errors, with API keys and tokens redacted. Click Delete next to the heading to clear it once you’ve fixed the problem. Errors are also written to the PHP error log with the prefix AI TTS:.

Related

• Fix audio generation errors
• How audio generation works