Comparison
Murmur vs Azure Speech
Azure Speech is a developer API from Microsoft. Murmur is a local voice studio for Mac and Windows. Here is how they differ on cost, privacy, cloning and output, using Azure Speech's own published details.
Free plan with Kokoro voices, 3,000 characters a day · 3-day Pro trial, then $49 once
Azure text to speech 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.
- azure text to speech
- microsoft azure tts
- Murmur vs Azure Speech
01 · Context
Quick verdict
Azure Speech is the better pick for enterprises and developers building voice-enabled apps, agents or accessibility features who need broad multilingual coverage, cloud-scale synthesis, and customizable brand or cloned voices backed by Azure security and compliance.
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
Azure Speech and Murmur side by side
Every Azure Speech entry below comes from Azure Speech's official website. Where the site did not say, the table says so instead of guessing.
| Azure Speech | Murmur | |
|---|---|---|
| Pricing model | Pay per use | One-time purchase |
| Free tier | Free (F0) tier includes "0.5 million characters free per month" for Neural text to speech (per-character billing). | Free plan and 3-day Pro trial; 7-day refund |
| Paid plans | Pay as you go: Billed per 1M characters; rates vary by voice type and region (see the Azure Speech pricing page) | $49 one-time for one computer (Mac or Windows), free plan and 3-day Pro trial, 7-day refund |
| Usage limits | Real-time synthesis via the Speech SDK or REST API truncates audio longer than 10 minutes; use the batch synthesis API for audio longer than 10 minutes (for example audiobooks or lectures). Billing is based on the total number of characters in each successfully processed request, including letters, numbers, spaces, tabs, punctuation and SSML markup (except the <speak> and <voice> tags); each Chinese character counts as two characters, including kanji, hanja and hanzi. | 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 | Depends on the voice: some local, some cloud | Local, on your Mac or PC |
| Works offline | Yes | Yes, after models are downloaded |
| Voice cloning | Yes | Yes, from a short reference recording |
| Languages | Standard neural voices available out of the box in 100+ languages and locales. | 600+ with the optional OmniVoice model |
| Export formats | MP3, WAV (RIFF PCM), OGG/Opus, Opus, WebM, RAW PCM, AMR-WB, G.722, TrueSilk, A-law/Mu-law | WAV and M4A on Mac; WAV on Windows |
| Commercial use | Available for commercial use under the terms governing your Azure subscription, including Microsoft's Service Specific Terms. Custom neural voice (voice cloning) is a Limited Access feature available by registration and approval only, restricted to Microsoft-approved use cases, with obligations to obtain explicit written permission from voice talent and to disclose the synthetic nature of the output. | Depends on the license of the model you generate with; check each model's license |
03 · Context
What Azure Speech does well
Broad language coverage: high-quality neural voices available out of the box in 100+ languages and locales, with multilingual voices that can switch languages via SSML.
Enterprise trust and compliance: built on Azure with 100+ compliance certifications and Microsoft's security investment, aimed at enterprise-grade, global-scale deployment.
Deep voice customization: custom neural voice (professional voice fine-tuning), personal voice, HD voices, and speaking styles/roles, plus visemes for facial animation.
Flexible delivery: real-time synthesis, asynchronous batch synthesis for long audio, and optional deployment in containers or embedded speech for edge and intermittent-connectivity scenarios.
04 · Highlights
When to stay with Azure Speech
| Decision factor | What it means |
|---|---|
| You need it on Web, API | Azure Speech 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 | Azure Speech 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 | Azure Speech 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: free (F0) tier includes "0.5 million characters free per month" for Neural text to speech (per-character billing). 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 | Azure Speech charges through pay per use. Murmur is $49 once with no character limits, so long audiobooks and repeated retakes cost nothing extra. |
| 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
- Azure Speech in Foundry Tools | Microsoft Azure checked 2026-10-14
- Pricing - Azure Speech in Foundry Tools | Microsoft Azure checked 2026-10-14
- Text to speech overview - Speech service - Foundry Tools | Microsoft Learn checked 2026-10-14
- Text to speech API reference (REST) - Speech service - Foundry Tools | Microsoft Learn checked 2026-10-14
- About the Speech SDK - Speech service - Foundry Tools | Microsoft Learn checked 2026-10-14
- Limited Access - Foundry Tools | Microsoft Learn checked 2026-10-14
FAQ
Common questions
Free (F0) tier includes "0.5 million characters free per month" for Neural text to speech (per-character billing). Paid plans: Pay as you go: Billed per 1M characters; rates vary by voice type and region (see the Azure Speech pricing page).
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.