IVR prompt generator
Type the text of a phone prompt, see exactly how its numbers, codes and dates will be read, then download it as 8 kHz audio for your phone system. English only for now.
Free to try, with limits: 400 characters per prompt and 10 prompts per hour. Audio is synthesised by Gradium TTS, so the text you enter is sent to Gradium to generate it. Do not enter real account numbers or personal details. See Privacy.
What gets rewritten
Ordinary text-to-speech reads 1222 as a quantity. A caller writing down a reference number needs it digit by digit. These patterns are rewritten before the text reaches the engine; anything the rules cannot place with confidence is left alone.
| Pattern | Written as | Read as |
|---|---|---|
| Booking and reference codes | A4-92-KJ7 | A, four, nine, two, K, J, seven |
| Account and phone numbers | 1-800-222-1222 | one, eight zero zero, two two two, one two two two |
| IBAN | FR76 3000 6000 … | F, R, seven six, three zero zero zero, six zero zero zero, … |
| Times | 14h30 | half past two in the afternoon |
| Dates | 03/04 | March fourth (or the third of April) |
Email addresses and URLs are passed through unchanged and handled by the engine itself. Amounts, decimals and four-digit numbers such as years are not touched. Dates are ambiguous: 03/04 is March 4 in the US and 3 April almost everywhere else, so you choose, except when the first number is above 12 and only one reading is possible.
Which format do I need?
Phone systems expect 8 kHz mono audio. The encoding depends on the carrier and the PBX.
| Format | Use it for |
|---|---|
| 8 kHz μ-law WAV | Twilio, Asterisk, most SIP trunks in the US |
| 8 kHz A-law WAV | Europe and other A-law carriers |
| 8 kHz 16-bit PCM WAV | Asterisk .wav, FreeSWITCH, general telephony |
| 24 kHz 16-bit PCM WAV | Web, video, anything that is not a phone line |
Not sure? Start with μ-law for a US carrier and A-law for Europe. Check your provider's documentation, because the two sound similar but are not interchangeable.
Limits of this page
This is a prompt generator, not a studio. It makes one short prompt at a time in six English voices. Other languages, voice cloning and batch export are not available yet; request access if you need them. IVR Voice Gen is built on Gradium TTS and is not affiliated with it.