Azure Text to Speech charges per character: $15 per million characters for prebuilt neural voices, and 500,000 characters a month free on the Free tier. Most people looking up Azure’s price are deciding which voice type to use, whether the free tier covers them, or whether to leave. This page covers both: Azure’s current pricing, free tier, and custom voice access first, then 10 alternatives and what to check before switching.
Azure Text to Speech pricing and free tier
Pay-as-you-go prices for the US East region, from Microsoft’s Azure Speech pricing page and Azure retail prices, checked October 5, 2026:
| Voice type | Price per 1M characters | Free tier (F0) per month |
|---|---|---|
| Neural (prebuilt voices) | $15 | 500K characters |
| Neural HD | $22 | Not included |
| Custom voice (professional), neural | $24 | Not included |
| Custom voice (professional), HD | $48 | Not included |
| Long audio (batch, prebuilt neural) | $100 | Not included |
A few things the table doesn’t show:
- The free tier doesn’t expire, but it’s small. 500,000 characters is about 11 hours of speech a month at a normal speaking rate, enough to prototype. Free resources are also rate-limited to 20 requests per minute, against 30 requests per second on Standard (Azure Speech quotas).
- Custom voices cost more than synthesis. Professional voice training bills $52 per compute hour, and each deployed custom voice endpoint bills $4.03 an hour while it’s hosted, whether or not it speaks.
- Custom voice needs approval. Professional voice and personal voice are limited access: you apply through Microsoft’s intake form, and the speaker records a consent statement before training.
- High volume gets a discount. Commitment tiers start at $960 a month for 80 million neural characters, which works out to $12 per million, with lower rates at 400 million and 2 billion characters.
- Prices vary by region. The figures above are for US East; check the pricing page for the region your resource runs in.
Azure’s prebuilt neural voices are competitively priced, and if you already run on Azure, the identity and billing integration is worth something. People usually look elsewhere for three reasons: they need lower latency for a live conversation, they need a cloned voice without an approval process, or they need voices that sound more natural on their own scripts.
An Azure text-to-speech migration starts with the voice, region, and speech controls your application uses. Compare alternatives against those requirements before comparing subscription prices.
What to compare
Azure speech services fit teams that already manage Azure resources. Cartesia provides Sonic through its own API. Switching providers changes authentication and request handling, and custom voice access needs its own review.
The options below cover different jobs; they are not a ranked benchmark. Check each provider’s current availability, features, and plan terms before committing.
Alternatives at a glance
| Product | Evaluate for | Check before choosing |
|---|---|---|
| Cartesia | Speech for voice applications | Model, language, and integration requirements |
| Speechify | Reading documents aloud | Reading app, studio, or API plan |
| Amazon Polly | TTS in AWS applications | Voice engine, region, and supported controls |
| Google Cloud Text-to-Speech | TTS in Google Cloud applications | Voice-specific features and quotas |
| IBM Watson Text to Speech | TTS API integration | Languages, controls, and service limits |
| Murf AI | Scripted voiceovers | Editing controls and export rights |
| ElevenLabs | Voice generation and dubbing | Model choice and usage limits |
| PlayHT | Existing TTS workflows | Current service and API availability |
| WellSaid Labs | Business voiceovers | Team workflow and language coverage |
| Synthesia | AI presenter videos | Avatar requirements and video exports |
Cartesia
We build Sonic, a text-to-speech model for voice applications. You can stream generated audio, select a voice, or use voice cloning with recordings you have permission to use. Try your own text in the playground before integrating the API.
For a complete voice agent, you also need speech recognition and conversation handling. Ink is our speech-to-text model, and Managed Agents is our voice agent builder. Sonic alone does not listen to callers or decide what to say.
Check language coverage, API requirements, and pricing for the model and plan you intend to use. Measure response time in your application: network travel and playback buffering contribute to what a user hears.
Speechify
Speechify has reading apps that turn documents and web pages into audio. If your goal is to listen to existing text, start with that workflow. Evaluate its reading, studio, and API products separately; access to one does not tell you what another includes.
Amazon Polly
Amazon Polly is AWS’s text-to-speech service. It is worth evaluating if your application already uses AWS identity and billing. Voice engines differ in language coverage and supported controls, so check the engine you plan to deploy rather than the service-wide feature list.
Google Cloud Text-to-Speech
Google Cloud Text-to-Speech provides speech synthesis through Google Cloud APIs. For an existing Google Cloud application, account and billing integration may simplify adoption. Confirm that your chosen voice supports the language, streaming behavior, and speech controls your application needs.
IBM Watson Text to Speech
IBM Watson Text to Speech is an API for generating speech from text. Include it if you are comparing speech services for an IBM-based deployment. Verify supported languages and pronunciation controls, then test output formats and service limits against your application.
Murf AI
Murf AI has a voiceover editor for working from scripts and matching narration to media. Consider it for training materials and recorded presentations. Test how much editing your script needs, and check whether the plan includes the exports and team access you need.
ElevenLabs
ElevenLabs has text-to-speech, voice cloning, and dubbing tools, as well as products for conversational applications. Compare the specific model and endpoint you would deploy. Test pronunciation and response time with your own scripts rather than treating every model as interchangeable.
PlayHT
PlayHT has been used for voice generation and TTS API integrations. Before building around it, confirm current service availability, API access, and support with the provider. If you are moving an existing integration, save the scripts and voice settings you need for comparison tests.
WellSaid Labs
WellSaid Labs focuses on voiceover production for business content, including training and internal communications. Evaluate how your team reviews scripts and handles pronunciation changes. Check language coverage and API access for the plan you intend to use.
Synthesia
Synthesia creates videos with AI presenters and narration. Evaluate it when the finished output needs a presenter on screen. For audio-only products, check whether a speech API would avoid paying for video features you do not use.
Test before you switch
List the SSML tags and pronunciation rules in your current requests. Check equivalents in the new API, then test audio encoding in your phone or browser client. Measure latency from the regions where your users connect, including peak concurrent traffic.
Compare cost using your expected usage and the features you need. Include retries, regenerated audio, concurrency limits, and commercial rights. A low starting price is not a workload estimate.