Paxa Labs and Amazon Polly for English, Thai, and Mandarin text to speech

A speech API for English, Thai, and Mandarin next to Amazon Polly, across Thai and Mandarin voices, word timing, streamed text, custom voices, and the cases each one suits.

If your product runs on Amazon Polly and now has to speak Thai, the voice list is the first thing to check. Polly's voice table covers 42 languages and variants, with one Mandarin voice and no Thai entry. Paxa TTS Flash v2 speaks English, Thai, and Mandarin, and all 50 of its voices speak all three. For English alone, Polly offers four engines and nine variants. There the voices on your own text and the rate at your volume decide it.

Sources:Amazon Polly voices(opens in a new tab)Amazon Polly languages(opens in a new tab)

Rates and claims last checked October 2026

Start free See full pricing

What differs

What differs
Dimension	Paxa Labs	Amazon Polly(opens in a new tab)
Thai voices	All 50 voices on Paxa TTS Flash v2 speak Thai, each with a published id and a fixed sample line. Paxa TTS Flash, model paxa-tts-flash-v1, carries 4 more in 3 regional registers.	Their voice table and language list carry no Thai entry.
Mandarin speech	All 50 voices speak Mandarin from Simplified Chinese text, in the same voice they use for English and Thai.	One Chinese (Mandarin) voice, Zhiyu, on their Neural and Standard engines.
English speech	All 50 voices speak English, billed at the same per-character rate as Thai and Mandarin.	Voices for nine English variants: US, British, Australian, Indian, Irish, New Zealand, Singaporean, South African, and Welsh. US English has voices on all four of their engines.
Word timing	Word, sentence, or utterance spans, returned with the audio on the same request and billed at 1.25 times the untimed rate.	Speech marks give word and sentence positions, plus visemes and SSML marks, on the Standard, Neural, and Long-Form engines. A speech marks request returns JSON in place of audio, and their pricing bills it by the character.
Text that arrives in pieces	A live WebSocket takes text fragments as they arrive and streams audio back. A cancel frame stops the reply when the listener talks over it.	StartSpeechSynthesisStream takes text incrementally over HTTP/2 and returns audio as it is synthesized. It runs on the Generative engine only, without speech marks.
Pronunciation and delivery	On Paxa TTS Flash v2, tags written into the text, such as [whispering] or [laughs], direct the delivery. A language field sets how codes, digits, and symbols are read.	SSML for phrasing, emphasis, and intonation, and custom lexicons, uploaded as XML, that change how particular words are pronounced.
Custom voices	No cloning. A voice can be designed from a written description in the console and called by its id on Paxa TTS Flash v2.	Brand Voice, a custom engagement with the Amazon Polly team that builds a voice for your organization's exclusive use.

Sources:Amazon Polly voices(opens in a new tab)Amazon Polly languages(opens in a new tab)Amazon Polly speech marks(opens in a new tab)Amazon Polly SynthesizeSpeech(opens in a new tab)Amazon Polly pricing(opens in a new tab)Amazon Polly bidirectional streaming(opens in a new tab)Amazon Polly features(opens in a new tab)

Paxa TTS Flash v2, per 1M characters: $10
Amazon Polly: $4-100
Their rate, as a multiple: 0.4-10×

Their range runs from the Standard voices to the Long-Form voices, the cheapest and the most expensive of their four engines, read from their own pricing page. Every rate is per one million characters of English text, counted in UTF-16 code units as ours is, with per-minute, per-token, per-byte, and per-Han-character rates converted to that unit.

What it costs at your volume

What it costs at your volume
Speech, per month	Paxa Labs	Amazon Polly	Difference
100,000About 111 minutes	Paxa Labs$1	Amazon Polly$0.40-10	Difference$0.60 more
1,000,000About 18 hours	Paxa Labs$10	Amazon Polly$4-100	Difference$6 more
5,000,000About 92 hours	Paxa Labs$50	Amazon Polly$20-500	Difference$30 more
20,000,000About 371 hours	Paxa Labs$200	Amazon Polly$80-2,000	Difference$120 more

Every figure is the two published rates multiplied by the volume beside it. Where the other vendor publishes a range, the low end is compared and the range is shown.

When Amazon Polly is the better choice

A team already on AWS adds speech inside the account and bill it already has. Polly's voice table covers 42 languages and variants, and we serve three languages. If your pipeline writes SSML or keeps pronunciation lexicons, Polly reads both as they stand. Speech marks with visemes suit avatars whose lips move with the audio. Brand Voice builds a voice for your organization alone. Choose us when your product speaks Thai, or when one voice has to speak English, Thai, and Mandarin.

Questions

Does Amazon Polly speak Thai?

Their voice table and their language list carry no Thai entry. For Mandarin, the voice table lists one voice, Zhiyu.

Can we keep Polly for other languages and use Paxa for English, Thai, and Mandarin?

Yes. Nothing in either API prevents routing requests by language. Thai text with English words in it goes to us as one request.

Do you publish quality or latency comparisons?

No. We publish no quality, accuracy, or latency figures of our own, for our model or anyone else's. Generate the same script on both and listen.

How current are the prices here?

They come from each vendor's published list rate, recorded in one place and re-checked on a fixed schedule. The month of the last check is at the top of the page.

Other comparisons

Start with free credits

Start free