Microsoft Azure Neural TTS

Azure text to speech is Microsoft's cloud speech synthesis service, part of Azure Speech within Foundry Tools (formerly Azure Cognitive Services). It converts text into natural speech using standard neural voices available out of the box in more than 100 languages and locales, plus higher quality HD voices, Azure OpenAI voices, and custom voice options that build a voice unique to a brand. It is a developer service rather than an editing app. Access is through the Speech SDK, a REST API and the Speech CLI, with a batch synthesis API for audio longer than 10 minutes such as audiobooks and lectures. SSML controls pitch, pauses, pronunciation, rate, volume and speaking style, standard voices are available at 24 kHz and 48 kHz, and viseme events for US English drive facial animation. Custom voice covers professional voice fine tuning and personal voice and is limited access, requiring an application. A no code Audio Content Creation tool in Speech Studio serves non developers, and text to speech avatars, prebuilt or custom, produce video. Pricing is pay as you go on characters processed, including spaces, punctuation and markup. The free F0 tier includes 0.5 million characters per month. Standard neural voices are $15 per 1 million characters. Custom professional voice synthesis is $24 per 1 million characters, or $48 for HD synthesis, with training at $52 per compute hour capped at $936 per training and endpoint hosting at $4.04 per model per hour. Commitment tiers run from $960 per month for 80 million characters to $15,000 for 2 billion. Some rows on the pricing page, including personal voice and certain avatar options, display no figure. The older cognitive services URL redirects; documentation and billing now sit under Foundry Tools.