GUIDE

An Azure text to speech alternative billed by time, not characters

Azure AI Speech has one of the widest choices of neural voices and locales, billed per character: $15 per million for neural voices and $22 for the HD ones at pay-as-you-go list prices, with a free tier of half a million characters a month. For an app that speaks for hours a day, the characters add up.

The SuperClonica speech API is billed by time, from $5 a month. Here is an honest comparison and how to move a request over.

Hear the voices free Or run the engine on your own server

The price of 216 hours

At 1,000 characters a minute, 216 hours of speech are about 13 million characters: about $194 with Azure neural voices and $285 with the HD ones at list prices. The SuperClonica Starter plan voices the same 216 hours for $5 a month when used without a break, 3 minutes of audio in every 10 minutes.

  • Free: 1 minute of audio a day.
  • Starter, $5: up to 216 hours a month.
  • Author, $10: up to 576 hours.
  • Pro, $20: up to 1,440 hours.
  • Studio, $50: up to 4,680 hours.

Voices: a catalog instead of locales

Azure organizes voices by locale. SuperClonica has a catalog of character, narrator, announcer and everyday voices made by authors, and a paid plan uses any of them through the API without buying it. A custom voice does not need an application form: you train it from your own recording on the site and use it like any other.

Move a request over

Instead of the Speech SDK or the Azure REST endpoint with an SSML body, post JSON to https://superclonica.com/api/v1/speech with Authorization: Bearer and your key: {"text": "...", "voice": "F1", "language": "en", "format": "mp3"}. Add "stream": true to play the audio as it comes; the first sound came from 300 ms in our measurements. A 429 or 503 answer carries Retry-After: wait that long and send the request again with the same Idempotency-Key.

When Azure is the better choice

Azure is ahead on languages and locales, including Chinese, on SSML with styles and visemes for lip sync, on regional hosting, containers for your own data center and enterprise contracts. If your volume fits half a million characters a month, its free tier may also cost nothing. Listen to both on your own text before deciding.