Amazon Polly alternative · Speech APIs

An Amazon Polly alternative for Hindi and Hinglish speech

Voice agents from ₹1 a minute at volume (₹1,00,000/month commitment), or ₹5 pay as you go, no commitment

₹1 is the T1 voice rate. On T1, Mirai phone calling (₹0.30/min) and call insights (₹0.20/min) are optional; T5 includes both. All prices exclude GST.

Mira TTS speaks Hindi, English and Hinglish at ₹1.80 per 1,000 characters through OpenAI's speech protocol. Amazon Polly lists two Hindi voices, Aditi (Standard) and Kajal (Neural), both bilingual in Hindi and Indian English, at $4 and $16 per million characters, with neural voices in the Mumbai region and a free tier. Choose Polly if you are on AWS and want a Mumbai endpoint and Speech Marks; choose Mira TTS for a voice built for code-switched Hindi calls and rupee billing.

No sign-up to look around.

Side by side

Amazon Polly and Voice Infra

Amazon Polly's column comes from Amazon Polly's own public pages, each cell linked to the page it was read from. Ours comes from our price list and docs. Pages change, so check the links before you decide.

Checked . Quotation marks are their words. Tell us if a cell is out of date.
QuestionAmazon PollyVoice Infra
Published price Standard $4.00, Neural $16.00, Generative $30 and Long-Form $100.00 per 1 million characters
Source: aws.amazon.com
₹1.80 per 1,000 characters, on pay as you go and on Business (₹50,000/month commitment), before GST
Hindi voices hi-IN: Aditi (Standard) and Kajal (Neural); no Generative or Long-Form Hindi voice listed
Source: docs.aws.amazon.com
Hindi voices that code-switch into English: ashu, neha, shruti and sameer
Hinglish Aditi and Kajal are bilingual Hindi and Indian English voices that can switch within one sentence, in Devanagari or Romanised Hindi
Source: docs.aws.amazon.com
Hindi, English and Hinglish in one voice; Hindi in Devanagari with English words in Latin script
Streaming and formats SynthesizeSpeech returns an audio stream: mp3, ogg_opus, ogg_vorbis, pcm (8 or 16 kHz), mulaw or alaw
Source: docs.aws.amazon.com
Raw PCM streams as it renders, 48 kHz by default; 8 kHz μ-law and A-law for phone lines; WAV files; a WebSocket for whole calls
API protocol Amazon Polly's SynthesizeSpeech API
Source: docs.aws.amazon.com
OpenAI's POST /v1/audio/speech with model mira-tts
Data location Neural voices supported in Asia Pacific (Mumbai), ap-south-1, among other regions
Source: docs.aws.amazon.com
Audio, speech recognition, TTS and the model run in India (Mumbai, Noida and Delhi). The control plane (workflow orchestration and the API) runs in Europe, off the audio path
Try it free 5M Standard characters a month; 1M Neural characters a month for 12 months; new customers get up to $200 in AWS Free Tier credits
Source: aws.amazon.com
No sign-up to look around the sandbox; ₹200 of credit once you confirm a work email; clips you play in the console are not charged

For six platforms in one table, see Voice AI platforms compared.

Which to pick

Which to pick

Choose Amazon Polly if…

  • You run on AWS and want the voice in the same account, with neural voices served from Asia Pacific (Mumbai), ap-south-1.
  • You want Speech Marks, the word and sentence timing metadata Polly sells alongside audio, for captions or lip sync.
  • You want a long free tier: 5 million Standard characters a month, and 1 million Neural characters a month for the first 12 months.
  • Your text is in Romanised Hindi. AWS says Aditi and Kajal read Hindi in Devanagari or Latin script and switch to Indian English within a sentence.

Choose Voice Infra if…

  • You want a choice of Hindi voices tuned for phone conversations, which code-switch into English the way callers do.
  • You want rupee billing on a prepaid wallet: ₹1.80 per 1,000 characters, before GST, with failed or partial requests not charged.
  • Your code already calls OpenAI's speech endpoint, or you use Pipecat. Ours takes the OpenAI request shape, and pipecat-mirai adds a ready TTS service.
  • You also need call transcription or a full phone agent from the same key and wallet.

In detail

An Amazon Polly alternative for Hindi and Hinglish speech: in detail

How do Amazon Polly and Mira TTS differ?

Polly is AWS’s speech synthesis service, with four voice engines on one price list. For Hindi it lists two voices: Aditi on the Standard engine and Kajal on the Neural engine. AWS describes both as fully bilingual in Hindi and Indian English, able to switch languages inside a sentence and to read Hindi in Devanagari or in Latin script. Neural voices run in the Mumbai region, and Polly’s Speech Marks return word and sentence timings for captions or lip sync.

Mira TTS is the voice our own phone agents use, offered on its own. It speaks Hindi, English and Hinglish with Hindi voices tuned for calls, and it is called through OpenAI’s speech protocol, so most frameworks that already speak to OpenAI can use it with a new base URL.

What does it cost?

Polly lists $4.00 per million characters for Standard voices and $16.00 for Neural, with Generative at $30 and Long-Form at $100.00, though neither of those last two has a Hindi voice on the voice list. The free tier covers 5 million Standard characters a month, and 1 million Neural characters a month for the first year.

Mira TTS is ₹1.80 per 1,000 characters on pay as you go and on Business, so 10 lakh characters cost 1,000 × ₹1.80 = ₹1,800 before GST. Requests that fail or stop part-way are free, and each response tells you the characters billed and the cost in paise.

What changes when you switch?

You replace a SynthesizeSpeech call with an OpenAI-style request: POST /v1/audio/speech with model: "mira-tts", a voice, a response_format and the text, up to 2,000 characters per request. Polly’s mulaw and alaw outputs map to our ulaw_8000 and alaw_8000, and its pcm maps to ours, which defaults to 48 kHz unless you set sample_rate. We have no Speech Marks equivalent, so keep Polly for any feature that depends on them.

For a live call, our WebSocket endpoint keeps one connection open, takes text as your model writes it, and lets you cancel a reply when the caller talks over it. On Pipecat, pipecat-mirai wraps that for you.

How should you compare them?

Take a script from your own calls and render it with Kajal and with two of our voices. Include names, amounts, dates and an OTP, and include lines that switch language mid-sentence, since both vendors say they handle that. Play the clips blind to people who speak Hindi every day, then play the best ones over a phone line. The Hindi TTS comparison lists the other options with their published prices.

If you are pricing a whole calling agent, our agents run from ₹1 a minute on T1 (₹1,00,000 monthly commitment) to ₹5 a minute pay as you go, with the voice included.

By Mirai Minds · Published

Questions

What do 10 lakh characters cost on each?

On Polly, 1,000,000 characters is $16.00 with Kajal (Neural) and $4.00 with Aditi (Standard), before the free tier. On Mira TTS it is 1,000 × ₹1.80 = ₹1,800 before GST. The currencies differ, so convert at your own rate.

Can Mira TTS read Romanised Hindi?

Write Hindi in Devanagari and keep English words in Latin script; that is how our examples are written and how we test the voices. If your source text is Romanised Hindi, test a sample before you switch, because Polly documents Romanised Hindi for Aditi and Kajal and our docs do not.

Can I get phone-ready audio from Mira TTS?

Yes. Ask for ulaw_8000 or alaw_8000 and forward each chunk to Twilio, Plivo or your SIP leg as it arrives. Exotel streams take pcm_8000.

Is Mira TTS ₹1 a minute?

No. ₹1 a minute is the T1 price of a whole phone agent with a ₹1,00,000 monthly commitment. Mira TTS on its own is priced by the character, at ₹1.80 per 1,000.

Check our column yourself.

Open the sandbox, build an agent and listen to it. Nobody asks for an email until you test a call.

Book a call

15 minutes with Sneh Mehta

Finding a time for us…

Open the sandbox