Speech

Speech to Text

Powered by the AT&T Watson℠ speech engine, the Speech to Text functionality supports speech-enabled apps that run on virtually any cellular or non-mobile network in the United States—it just needs an internet connection. There are nine speech contexts available that are built and maintained by AT&T, and we'll continue to tune them for you. You send us audio, and we'll send you text of what your end user said. It's that easy. Then you can perform any action you want with the text. Choose the context that best fits your app:

• Gaming
• Social Media
• Web Search
• Business Search
• Voicemail to Text
• SMS
• Question and Answer
• TV (with the option of getting AT&T U-verse Electronic Program Guide information in the response. e.g. showtimes, etc. based on location)
• Generic—A general purpose context that can automatically detect and transcribe the English and Spanish languages.

Environments Supported

*Android support for Speech to Text Custom is coming soon.

Key Features

• Cross-carrier
• Native and HTML5-based SDKs
• Nine optimized speech contexts
• Now with acoustic models optimized for smartphones and limited distance (where users are not speaking right in front of and directly into a device).

How Can I Use Speech to Text?

	Gaming Allow your gamers to use their voice to get to the next level. We've optimized this context for: gamer-to-gamer communication, war games, role play, alphanumeric input, sports, and more.
	Family Communication Use the Voicemail to Text context to transcribe family voicemails to text, then post the text to a centralized area for all family members to see.
	TV—Programming and Remote Control Use the TV context to turn your user's mobile phone into a remote control. This context will allow your users to search for their favorite TV shows, movies, actors, etc. using their voice.
	Local Points of Interest Use the Business Search context to transcribe their users' local searches into text. Then have your app send the text to your favorite local search engine to get the results of their search. Now you have the option to get AT&T U-verse Electronic Program Guide results, as well.
	Education Develop an app that helps kids with their homework and research projects? Want to have web searches fulfilled by a specific (kid-friendly) search engine? Try the Web Search context.

Text to Speech

This one speaks for itself.

Just send us text as part of the API call, and one of our polite, well-versed characters will be happy to 'say' what you sent. Our current offer includes:

• US English (Crystal or Mike)
• US Spanish (Rosa or Alberto)

Powerful, optional controls include:

• Tempo control (speed at which the audio is 'read' by the character)
• SSML support
• WAV, AMR, and AMR Wideband audio formats
• Volume control

Key Features

• Cross-carrier
• HTML5-based SDKs
• US English and Spanish-speaking characters available

How Can I Use Text to Speech?

	Tell Me What I Said Whether you are enabling a hands-free in-car experience, or just letting people 'say' instead of 'type'...you can use Text to Speech as a way to confirm to the end user that your app transcribed what they said correctly...and giving the option to confirm or edit.
	Read It To Me Text to Speech can be used to read the content of email and SMS messages back to users who are not able to read a message at any given time. e.g. driving or visually impaired.

Speech to Text Custom

You now have 2 powerful tools for customizing the Speech API for your own app:

• Grammar List
  • Allows you to send a list of words or phrases that your end users say along with the audio to be transcribed to text.
  • Perfect for command / response applications, or those in which end users only say a limited set of words or phrases
  • Supports 19 languages, including: Spanish, Mandarin, Japanese, French, Italian and German
  • Allows you to send pronunciation rules, so important words are pronounced correctly
  • Note: This context limits recognition to the words uploaded in the list.
• Generic with Hints
  • Similar to Grammar List, you can send a list of words or phrases that your end users say.
  • In addition, if end users say words outside the list, this context will also attempt to recognize the words using the Generic Speech to Text context.
  • Similar to the Generic Speech to Text context, English and Spanish languages are supported.

Key Features

• App-specific tuning
• Cross-carrier
• Native and HTML5-based SDKs
• Generic with Hints supports English and Spanish
• Grammar List supports 19 languages
• Acoustic models optimized for smartphones and limited distance (where users are not speaking right in front of and directly into a device).

How Can I Use Speech to Text?

	I Know What I Want Allow your 'regulars' to be able to quickly order what they want, without having to scroll through a list and finding what they want. Send your menu items or inventory list along with the audio, and allow our Speech API to understand and transcribe your unique products. The Grammar List can do this for you in 19 languages.
	Products and More A company uses Generic with Hints context to improve the recognition of company-specific jargon in its internal message tool. Employees say a wide range of vocabulary when messaging, and the company wants to be sure that their name, products and other key words are transcribed correctly.

Want more details? Visit our Docs section.

Sample Demo