IBM Research

IBM Text-to-Speech: Unconstrained U.S. English Text Demo This page demonstrates some of our work in unconstrained text-to-speech research. Our system generation process is automatic and can be built for any custom voice. From each voice recording, small segments of speech are extracted and stored in our speech database. During synthesis, these speech segments are assembled using specially crafted algorithms to generate any word in the language. Type in your text below and check it out. IBM's commercial TTS site contains more access points to available IBM products and product information. To hear synthesized speech, type in text (30-words or less) and press "Speak" - audio player will be launched*:    Select Voice: US Female 1 (22K) US Male 1 (22K) US Female 2 (22K) US Male 2 (22K) Comments If you have a comment on the quality of the generated voice sample, we'd like to hear from you: If you would like a reply, please provide an email address where we can contact you (OPTIONAL): This page is to be used for demonstration purposes only, usage will be limited to approximately 30 accesses per site per day. This application uses a flexible architecture wherein audio data is returned as a response to this page. Certain browser-plugin combinations may have problems with this response; in such cases we recommend using the "Download" button to save and play the file locally. Click here for International language examples

About your Privacy: The information collected here will not be used for any future marketing or promotional contacts beyond the scope of this transaction. The fields indicated with an asterisk are required to complete this transaction; other fields are optional. If you don't wish to provide us with required information, please use the back button on your browser to return to the previous page. IBM Privacy Policy