Comparison
Murmur vs Amazon Polly
Amazon Polly is a developer API from Amazon Web Services. Murmur is a local voice studio for Mac and Windows. Here is how they differ on cost, privacy, cloning and output, using Amazon Polly's own published details.
Free plan with Kokoro voices, 3,000 characters a day · 3-day Pro trial, then $49 once
Amazon Polly alternative
Murmur runs locally on Apple Silicon and Intel Macs and on 64-bit Windows 10 and 11 as a beta. The Mac app adds projects, a batch queue, community voices, and automation.
- amazon polly text to speech
- amazon polly pricing
- Murmur vs Amazon Polly
01 · Context
Quick verdict
Amazon Polly is the better pick for developers and businesses that need to add scalable, pay-per-character cloud text-to-speech to applications, contact centers, and media pipelines via an API rather than a consumer app.
Murmur is the better pick if you work on a Mac or PC, want speech generated locally, and would rather pay $49 once than track characters, credits or a subscription.
02 · Overview
Amazon Polly and Murmur side by side
Every Amazon Polly entry below comes from Amazon Polly's official website. Where the site did not say, the table says so instead of guessing.
| Amazon Polly | Murmur | |
|---|---|---|
| Pricing model | Pay per use | One-time purchase |
| Free tier | Upon sign-up, new customers can synthesize millions of characters for free each month for the first 12 months: 5 million characters per month for Standard voices, 1 million characters per month for Neural voices, 500 thousand characters per month for Long-Form voices, and 100 thousand characters per month for Generative voices (the first 12 months). | Free plan and 3-day Pro trial; 7-day refund |
| Paid plans | Standard voices (pay-as-you-go): $4.00 per 1 million characters for speech or Speech Marks requests (when outside the free tier); Neural voices (pay-as-you-go): $16.00 per 1 million characters for speech or Speech Marks requested (when outside the free tier); Long-Form voices (pay-as-you-go): $100.00 per 1 million characters for speech or Speech Marks requested (when outside the free tier) | $49 one-time for one computer (Mac or Windows), free plan and 3-day Pro trial, 7-day refund |
| Usage limits | Free tier allowances are per month (Standard 5 million characters; Neural 1 million; Long-Form 500 thousand; Generative 100 thousand for the first 12 months). Billing is monthly based on characters processed, and requests for both speech and Speech Marks are charged based on characters sent. AWS maintains default service limits on throttling, operations, and SSML use as described in 'Limits in Amazon Polly'. | No character, minute or credit limits |
| Platforms | Web, API | Mac (Apple Silicon or Intel) and Windows 10/11, 64-bit (beta) |
| Where speech is generated | In the vendor's cloud | Local, on your Mac or PC |
| Works offline | No | Yes, after models are downloaded |
| Voice cloning | Not stated on the official site | Yes, from a short reference recording |
| Languages | 100+ male and female voices in 40+ language and language variants | 600+ with the optional OmniVoice model |
| Export formats | MP3, OGG (Vorbis), PCM | WAV and M4A on Mac; WAV on Windows |
| Commercial use | As between you and AWS, your Polly output belongs to you; the service does not restrict storing and reusing generated speech, and there are no additional costs for generating static voice prompts that are replayed multiple times or content used in mass notification systems. You must own the rights to any third-party text you input. | Depends on the license of the model you generate with; check each model's license |
03 · Context
What Amazon Polly does well
Large voice catalog: 100+ lifelike male and female voices across 40+ languages and language variants, including Standard, Neural, Long-Form, and Generative voice engines.
Pay-as-you-go pricing charged only for characters converted to speech or Speech Marks, with free caching and replay of generated audio at no additional cost.
Rich controls via W3C SSML (pitch, rate, volume, whispering, Newscaster style, time-driven prosody) plus custom lexicons for pronunciation of names and acronyms.
Speech Marks metadata enables synchronized visuals such as karaoke-style word highlighting and lip-synced avatar animations.
04 · Highlights
When to stay with Amazon Polly
| Decision factor | What it means |
|---|---|
| You need it on Web, API | Amazon Polly lists Web, API support. Murmur runs as a desktop app on Mac and Windows, so it cannot follow you to a browser, phone, or tablet. |
| You are generating speech from your own software | Amazon Polly offers programmatic access for apps and servers. Murmur has a local CLI and MCP server for automation on one Mac, which is not a hosted API for your customers. |
| You want nothing to install or maintain | Amazon Polly runs its voices on the vendor's servers, so there are no model downloads and no dependence on your computer's memory. |
| Your needs fit inside the free tier | The official free tier: upon sign-up, new customers can synthesize millions of characters for free each month for the first 12 months: 5 million characters per month for Standard voices, 1 million characters per month for Neural voices, 500 thousand characters per month for Long-Form voices, and 100 thousand characters per month for Generative voices (the first 12 months). If that covers you, there is no reason to pay for anything. |
05 · Highlights
When Murmur is the better fit
| Decision factor | What it means |
|---|---|
| Your scripts are private or unreleased | Murmur generates on your Mac or PC, so client scripts, manuscripts and drafts are not uploaded for synthesis. |
| You generate a lot of audio | Amazon Polly charges through pay per use. Murmur is $49 once with no character limits, so long audiobooks and repeated retakes cost nothing extra. |
| You want your own voice or a designed voice | The official Amazon Polly site does not list voice cloning. Murmur clones from a short reference recording and can design new voices from a text description. |
| You produce finished audio, not just listening | Murmur is built around finished audio. The Mac app adds multi-speaker projects, per-line retakes and WAV or M4A export for editors such as Final Cut Pro, iMovie and Logic. The Windows beta exports WAV. |
Sources
Where these facts come from
- Amazon Polly - AI Voice Generator checked 2026-10-14
- Amazon Polly Pricing checked 2026-10-14
- Amazon Polly Features checked 2026-10-14
- Amazon Polly FAQs checked 2026-10-14
FAQ
Common questions
Upon sign-up, new customers can synthesize millions of characters for free each month for the first 12 months: 5 million characters per month for Standard voices, 1 million characters per month for Neural voices, 500 thousand characters per month for Long-Form voices, and 100 thousand characters per month for Generative voices (the first 12 months). Paid plans: Standard voices (pay-as-you-go): $4.00 per 1 million characters for speech or Speech Marks requests (when outside the free tier); Neural voices (pay-as-you-go): $16.00 per 1 million characters for speech or Speech Marks requested (when outside the free tier); Long-Form voices (pay-as-you-go): $100.00 per 1 million characters for speech or Speech Marks requested (when outside the free tier).
macOS · Local TTS
Generate speech on your own computer
Murmur is $49 one-time for one computer, Mac or Windows, with a free plan and a 3-day Pro trial. Purchases have a 7-day refund window. Listen to the public samples before you buy.