Cache audio — never synthesise twice
TTS bills per character, so re-generating the same phrase is pure waste. Cache and reuse clips, and match the voice tier to the need. For the reverse direction see the speech-to-text calculator.
How this calculator works
The TTS Price Calculator estimates your monthly text-to-speech API bill by multiplying two inputs: the number of characters synthesized per month and the price per one million characters charged by your provider. Because TTS APIs bill per character of input text, your total cost scales directly and linearly with volume. The two main drivers are therefore straightforward: how much text you convert to audio each month, and the per-character rate you pay. Entering both figures returns a projected monthly spend, letting you compare providers or forecast the effect of higher usage before you commit.
The key trade-off to watch is that character count, not audio length or word count, drives the bill. Spaces, punctuation, and markup tags like SSML all count toward your total, so verbose or heavily formatted text costs more than the spoken output might suggest. A small difference in the per-million rate also compounds quickly at scale, so it is worth caching or reusing frequently generated audio and trimming unnecessary characters from your input. Running a few volume scenarios through the calculator helps you spot where costs accelerate and choose a pricing tier that matches your actual usage.
Frequently asked questions
How is text-to-speech priced?
Almost always per character of input text, quoted per 1 million characters. Standard voices are cheapest; high-quality neural, expressive and cloned voices cost several times more per character.
What drives TTS cost up?
Volume and voice tier. Narrating long content (audiobooks, courses) burns millions of characters, and premium voices multiply the rate. Cache generated audio so you never pay to synthesise the same text twice.