What different audio volumes cost
Monthly cost at the default speaking rate, for a small app, a podcast or course library, and a large narration pipeline.
| Model | 10 h/mo | 100 h/mo | 1,000 h/mo |
|---|---|---|---|
| Amazon Polly Standard | $1.73 | $17.28 | $173 |
| Google Cloud TTS Standard | $1.73 | $17.28 | $173 |
| Google Cloud TTS WaveNet | $1.73 | $17.28 | $173 |
| Gemini 3.8 Flash-Lite TTS | $5.45 | $54.54 | $545 |
| Voxtral TTS 3 | $5.62 | $56.16 | $562 |
| Azure AI Speech Neural TTS | $6.48 | $64.80 | $648 |
| TTS-1 | $6.48 | $64.80 | $648 |
| xAI Text to Speech | $6.48 | $64.80 | $648 |
| Amazon Polly Neural | $6.91 | $69.12 | $691 |
| Google Cloud TTS Neural2 | $6.91 | $69.12 | $691 |
| Gemini 3.8 Flash TTS | $8.15 | $81.54 | $815 |
| Azure AI Speech Neural HD TTS | $9.50 | $95.04 | $950 |
| Amazon Polly Generative | $12.96 | $130 | $1,296 |
| Google Cloud TTS Chirp 3: HD | $12.96 | $130 | $1,296 |
| TTS-1 HD | $12.96 | $130 | $1,296 |
| Amazon Polly Long-Form | $43.20 | $432 | $4,320 |
| Google Cloud TTS Studio | $69.12 | $691 | $6,912 |
Character billing vs token billing
Most speech APIs bill the characters you send, so cost depends only on text length. Gemini TTS bills tokens instead: a small charge for the input text plus audio output tokens at 25 per second of generated speech. That makes Gemini's cost follow audio length, so slow narration with long pauses costs more per character than fast speech. For Gemini-only budgets by minute and hour, including Batch, Flex and Priority modes, use the Gemini 3.8 TTS Cost Calculator.
How to choose a voice tier
- Standard voices are the cheapest and fine for alerts, IVR prompts and accessibility read-outs.
- Neural voices sound natural enough for most apps and cost about four times as much.
- HD, generative and studio-grade voices suit podcasts, audiobooks and ads, at several times the neural price.
- Check language and voice coverage first; the cheapest tier may not offer your language.
Related pricing pages
Provider pages list every model with its sources: OpenAI, Google Gemini, Mistral AI and xAI. Embedding prices are compared on the embedding API pricing page. The raw records are in the AI API Pricing Dataset.
Frequently asked questions
Which text-to-speech API is the cheapest?
Among models not scheduled to shut down, Amazon Amazon Polly Standard has the lowest rate here at about $4.00 per 1M characters (about $0.17 per hour of audio). The most expensive is Google Cloud Google Cloud TTS Studio at $160 per 1M characters. Voice quality and language coverage differ a lot between tiers, so listen to samples before choosing on price.
How much does an hour of TTS audio cost?
It depends on how many characters you speak per minute. At 720 characters per minute (AWS's own example of 1 million characters for about 23 hours of speech), one hour is 43,200 characters, so a $16 per 1M character voice costs about $0.69 per hour. Change the speaking rate in the calculator to match your content.
How do Amazon Polly, Google Cloud and Azure neural voices compare on price?
Per 1M characters: Amazon Amazon Polly Neural $16.00, Google Cloud Google Cloud TTS Neural2 $16.00, Microsoft Azure Azure AI Speech Neural TTS $15.00. All three also sell cheaper standard voices and pricier premium tiers, which are listed in the table.
Is OpenAI's TTS API being shut down?
OpenAI has deprecated TTS-1 and TTS-1 HD and scheduled their shutdown for 2027-01-06. Until then they stay billed at $15.00 per 1M characters (TTS-1) and $30.00 per 1M characters (TTS-1 HD). OpenAI points users to gpt-realtime-2.1-mini through the Realtime API, which is token-billed and not a drop-in replacement for the Speech endpoint.
Why does Gemini TTS show an approximate price per 1M characters?
Gemini TTS bills tokens, not characters: text input tokens plus audio output tokens at Google's documented 25 tokens per second. The per-character figure converts that with the calculator's speaking-rate assumption. Google has published higher rates from 2027-01-01: Gemini 3.8 Flash-Lite TTS $1 text + $12 audio / 1M tokens; Gemini 3.8 Flash TTS $1 text + $18 audio / 1M tokens.
Do these prices include free tiers?
No. Amazon Polly and Google Cloud Text-to-Speech include monthly free character allowances, and Azure has a free F0 tier. The table shows the paid rate you pay after any free usage, so small projects can cost less than the calculator shows.
Where do these prices come from?
Each price is a verified record in the AICostBudget pricing dataset, linked to the provider's official pricing page or price list with its verification date. Amazon and Azure prices are the public on-demand rates shared by their commercial regions; government clouds are excluded.