Best for: Developers building custom voices across broad language coverage.
- Text-to-speech API buyer fit
- Broad language coverage and custom voices
Categories › AI Text-to-Speech Software
Best overall: Azure AI Speech is the pick for broad language coverage and custom voices, with Neural voices at $15 per 1 million characters and 0.5 million characters free monthly. Best value: Amazon Polly is the pick for standard AWS synthesis, at $4 per 1 million characters with 5 million Standard characters free each month. Google Cloud Text-to-Speech is the pick for low-latency streaming; Deepgram is the pick for real-time voice agents with responses in as low as 80 ms.
Updated
Best for: Developers building custom voices across broad language coverage.
Best for: Developers using AWS apps billed by character.
Best for: Developers streaming Chirp 3: HD voices.
Best for: Developers needing expressive, low-latency models.
Best for: Developers building real-time voice agents.
Best for: Voice-agent developers needing sub-100 ms audio.
| Product | Text-to-speech API buyer fit |
|---|---|
| Azure AI Speech | Broad language coverage and custom voices[1] [2] [3] |
| Amazon Polly | Apps on AWS[4] [5] [6] [7] [8] |
| Google Cloud Text-to-Speech | Streaming Chirp 3: HD voices[9] [10] [11] |
| ElevenLabs | Expressive voices and low-latency models[12] [13] [14] [15] [16] |
| Deepgram | Real-time voice agents[17] [18] |
| Murf | Voice agents needing sub-100 ms audio[19] [20] [21] |
Neural/Neural HD Flash (real-time and batch): $15 per 1M characters
Speech Synthesis Markup Language (SSML) is an XML-based markup language used to customize text to speech
Multiply the result by the unit price of $15 per million characters to estimate the monthly cost.
Amazon Polly’s Standard voices are priced at $4.00 per 1 million characters
Use SSML, a W3C standard XML-based markup language, to adjust phrasing, emphasis, and intonation
Amazon Polly’s Standard voices are priced at $4.00 per 1 million characters for speech or Speech Marks requests (when outside the free tier). For Amazon Polly’s Standard voices, the free tier includes 5 million characters per month for speech or Speech Marks requests.
The following languages are supported by Amazon Polly and can be used to synthesize speech.
24 | German (Austrian) | de-AT | Hannah | Female | Yes | No | Yes | No 25 | German (Swiss) | de-CH | Sabrina | Female | Yes | No | Yes | No
Chirp 3: HD voices (sku:F977-2280-6F1B) 0 to 1 million characters US$0.00003 per character (US$30 per 1 million characters)
These voices offer high-quality audio, low-latency streaming
Text-to-Speech is priced based on the number of characters sent to the service to be synthesized into audio each month.
v4 Text to Speech New $0.08 Price per 1K characters
Flash v2.5: 32 languages, ~75 ms model latency, up to 40,000 characters per request
Stream real-time audio from text
Access 17,500+ voices in 90+ languages.
Starter $6 per month
Nova models support 60+ languages, with capabilities including speaker diarization, Smart Formatting, and Keyterm Prompting.
Usage-based rates across every product, billed per second of audio, per 1,000 characters, or per minute of connection time.
Optimized for conversational AI and real-time voice agents with time-to-first-audio under 100ms $0.01 / 1000 characters
Pay as you go: With our pay-as-you-go plan, you can purchase 1000 characters at $0.03 (minimum purchase amount is $2).
Falcon 2.1 costs 1 cent per 1000 characters.
| Product | Best for | Published price |
|---|---|---|
| Azure AI Speech [vendor source] | Developers building custom voices across broad language coverage [vendor source] | $15 per 1M characters (Neural); 0.5M free a month [vendor source] |
| Amazon Polly [vendor source] | Developers using AWS apps billed by character [vendor source] | Standard $4, Neural $16, Generative $30 per 1M characters [vendor source] |
| Google Cloud Text-to-Speech [vendor source] | Developers streaming Chirp 3: HD voices [vendor source] | Chirp 3: HD US$30 per 1M characters; Standard US$4 [vendor source] |
| ElevenLabs [vendor source] | Developers needing expressive, low-latency models [vendor source] | v4 $0.08 per 1K characters; Flash $0.04 [vendor source] |
| Deepgram [vendor source] | Developers building real-time voice agents [vendor source] | Aura-2 $0.030 per 1,000 characters; Flux TTS $0.045 [vendor source] |
| Murf [vendor source] | Voice-agent developers needing sub-100 ms audio [vendor source] | API: $0.03 per 1,000 characters (studio voices); $0.01 (voice-agent model) [vendor source] |
Azure AI Speech bills per character: Neural and Neural HD Flash voices cost $15 per 1M characters, with 0.5 million characters free per month. Voices cover 100+ languages and locales, SSML supports lexicons and speaking styles, and custom voices can be built with professional voice fine-tuning.
Amazon Polly charges by characters processed. Standard costs $4, Neural $16, Generative $30 and Long-Form $100 per 1 million characters. Free tiers cover 5 million Standard characters a month, and 1 million Neural characters a month for the first 12 months. It supports SSML and custom lexicons.
Google lists Chirp 3: HD voices with low-latency streaming at US$30 per 1 million characters after 1 million free; Standard voices are US$4 and Neural2 US$16 per 1 million. SSML tags control pauses, number formats and pronunciation, and instant custom voice costs US$60 per 1 million characters.
ElevenLabs offers streaming text-to-speech through an API. Flash/Turbo costs $0.04 per 1,000 characters (32 languages, up to 40,000 characters per request, ~75 ms model latency); v4 costs $0.08 per 1,000 characters and supports 90+ languages. Audio streams in Opus at 48 kHz, with PCM for post-production.
Deepgram Aura-2 responses start in as low as 80 ms. Aura-2 costs $0.030 per 1,000 characters, Aura-1 $0.015, and Flux TTS $0.045 on pay-as-you-go. Tone, pacing and pronunciation carry across turns, and Flux TTS ships in English plus 10 additional languages. Deepgram lists a $200 free credit before pay-as-you-go.
Murf's API is $0.01 per 1,000 characters for its voice-agent model with time-to-first-audio under 100 ms, and $0.03 per 1,000 characters for studio-quality speech.
For speech-to-text and live meeting notes, see the AI meeting assistant category.
Google Cloud Text-to-Speech lists low-latency streaming and SSML, Amazon Polly supports SSML and lexicons, and Azure AI Speech supports SSML with lexicons.
Amazon Polly Standard is $4 per 1 million characters, Azure Neural $15, Google Chirp 3: HD US$30 and ElevenLabs v4 $0.08 per 1,000 characters.
Azure includes 0.5 million characters free a month, Amazon Polly 5 million Standard characters a month, Google 1 million Chirp 3: HD characters, and Deepgram a $200 credit.
Deepgram is the pick, with responses in as low as 80 ms and Aura-2 at $0.030 per 1,000 characters.
Ranked on documented capabilities and verified pricing from official vendor sources. Every fact links to its source and check date, and the criteria are specific to this buying situation.
Matrix comparison for AI text-to-speech buyers. Criteria cover pricing basis, free limits, language and accent coverage, commercial-use terms, pronunciation and SSML controls, export formats and integrations, APIs and streaming, vendor-stated quality, security, and each page’s buying situation. Vendor facts link to primary pages with UTC retrieval times and verbatim excerpts. Prices retain the vendor’s currency and stated billing basis. Ranked on documented capabilities and verified pricing from official vendor sources.
Want your software in this comparison? A vendor profile listing costs $9.90 via Stripe and goes live within 24 hours of payment. Comparison-list placements cost a fee, and links are dofollow. Factual corrections are free.
Submit your software →Send a correction with a source link and we review it against the vendor's own documentation.
Suggest a correction