What is neural text-to-speech?
Microsoft neural text-to-speech uses deep neural networks to make the voices of computers nearly indistinguishable from recordings of people. With the human-like natural prosody and clear articulation of words, neural text-to-speech has significantly reduced listening fatigue when you interact with AI systems.
The patterns of stress and intonation in spoken language are called prosody. Traditional text-to-speech systems break down prosody into separate linguistic analysis and acoustic prediction steps that are governed by independent models. That can result in muffled, buzzy voice synthesis. Microsoft neural text-to-speech capability does prosody prediction and voice synthesis simultaneously, uses deep neural networks to overcome the limits of traditional text-to-speech systems in matching the patterns of stress and intonation in spoken language, and synthesizes the units of speech into a computer voice. The result is a more fluid and natural-sounding voice.
In this overview, you learn about the benefits and capabilities of the text-to-speech service, which enables your applications, tools, or devices to convert text into human-like synthesized speech. Use human-like neural voices, or create a custom voice unique to your product or brand. For a full list of supported voices, languages, and locales, see supported languages.
This documentation contains the following article types:
- Quickstarts are getting-started instructions to guide you through making requests to the service.
- How-to guides contain instructions for using the service in more specific or customized ways.
- Concepts provide in-depth explanations of the service functionality and features.
- Tutorials are longer guides that show you how to use the service as a component in broader business solutions.
Bing Speech was decommissioned on October 15, 2019. If your applications, tools, or products are using the Bing Speech APIs or Custom Speech, we've created guides to help you migrate to the Speech service.
Asynchronous synthesis of long audio - Use the Long Audio API to asynchronously synthesize text-to-speech files longer than 10 minutes (for example audio books or lectures). Unlike synthesis performed using the Speech SDK or speech-to-text REST API, responses aren't returned in real time. The expectation is that requests are sent asynchronously, responses are polled for, and that the synthesized audio is downloaded when made available from the service.
Platform neural voices - Deep neural networks are used to overcome the limits of traditional speech synthesis with regard to stress and intonation in spoken language. Prosody prediction and voice synthesis are performed simultaneously, which results in more fluid and natural-sounding outputs. Neural voices can be used to make interactions with chatbots and voice assistants more natural and engaging, convert digital texts such as e-books into audiobooks, and enhance in-car navigation systems. With the human-like natural prosody and clear articulation of words, neural voices significantly reduce listening fatigue when you interact with AI systems. For a full list of platform neural voices, see supported languages.
Fine-tune TTS output with SSML - Speech Synthesis Markup Language (SSML) is an XML-based markup language used to customize text-to-speech outputs. With SSML, you can not only adjust pitch, add pauses, improve pronunciation, change speaking rate, adjust volume, and attribute multiple voices to a single document, but also define your own lexicons or switch to different speaking styles. With the multi-lingual voices, you can also adjust the speaking languages via SSML. See how to use SSML to fine-tune the voice output for your scenario.
Visemes - Visemes are the key poses in observed speech, including the position of the lips, jaw and tongue when producing a particular phoneme. Visemes have a strong correlation with voices and phonemes. Using viseme events in Speech SDK, you can generate facial animation data, which can be used to animate faces in lip-reading communication, education, entertainment, and customer service. Viseme is currently only supported for the
en-USEnglish (United States) neural voices.
Sample code for text-to-speech is available on GitHub. These samples cover text-to-speech conversion in most popular programming languages.
In addition to neural voices, you can create and fine-tune custom voices unique to your product or brand. All it takes to get started are a handful of audio files and the associated transcriptions. For more information, see Get started with Custom Neural Voice
When using the text-to-speech service, you are billed for each character that is converted to speech, including punctuation. While the SSML document itself is not billable, optional elements that are used to adjust how the text is converted to speech, like phonemes and pitch, are counted as billable characters. Here's a list of what's billable:
- Text passed to the text-to-speech service in the SSML body of the request
- All markup within the text field of the request body in the SSML format, except for
- Letters, punctuation, spaces, tabs, markup, and all white-space characters
- Every code point defined in Unicode
For detailed information, see Pricing.
Each Chinese, Japanese, and Korean language character is counted as two characters for billing.
Migrate to Neural Voice
We are retiring the standard voices on 31st August 2024 and they will no longer be supported after that date. The announcement has been sent out to all existing Speech subscriptions before 31st August 2021. During the retiring period (31st August 2021- 31st August 2024), existing standard voice users can continue to use their standard voices, all new users/new speech resources should move to the neural voices.
- Review the price structure and listen to the neural voice samples at the bottom of the page to determine the right voice for your business needs.
- To make the change, follow the sample code to update the voice name in your speech synthesis request to the supported neural voice names in chosen languages by 31 August 2024. Starting 1st September 2024, standard voices will no longer be supported, please use neural voices for your speech synthesis request, on cloud or on prem. For on-prem container, please use the neural voice containers and follow the instructions.