Phone Notify! Text-to-Speech Best Practices

Text-To-Speech (TTS) is a variation of speech synthesis that turns text into spoken words. CDYNE’s text-to-speech engine converts raw text to written-out words, and then assigns phonetic transcriptions to those words. The text is then read to the recipient using a computer generated voice. Text-to-speech can also be combined with recorded .wav files to create relevant call scripts according to your phone messaging web service needs.

With the Phone Notify! call script, you can customize the text-to-speech portion in many different ways.  Whether you need your call recipient to hear individual letters, URLs, or alphanumerically spoken letters, this can be accomplished with minimal effort in addition to adjusting the voice used within your call script to suit your needs.

The console application example below was created within Visual Studio using a C# .Net environment and utilizes the NotifyPhoneBasic method of the API.

1. Set Up the Application

2. Set Up the Method

To begin utilizing various features available to customize the text-to-speech portion of your call script, you will need to setup the method as shown below.

3. Customize Call Script

Once your method has been setup you can then begin customizing your call script.  In the call script below, several commands are being used to adjust the speech rate, pitch, and pronunciation of an individual letters within an URL.

The advanced command, ~WaitForSilence()~ tells the script to wait for half of a second of silence before beginning the script.  To adjust the pitch of the text-to-speech voice you will need to use <prosody pitch=’x-low’> </prosody>.  If you find that a voice volume is too low or too high, you would use the <prosody volume=’x-soft’> </prosody> speech command.  When providing individual letters or URLs within your call script, you may find that the letters or URL is spoken too slow.  To adjust the speed at which words, letters, or URLs are spoken within your call script you will use <prosody rate=’x-slow’> </prosody>.   If you would like to add pauses between letters, you would use <break strength=’weak’ />. When using the above mentioned advanced speech control commands within the call script, your call script will sound less robotic and recipients are less likely to notice they have received an automated call from your company.

If you would like for individual letters to be spoken alphanumerically, you would simply change this line within your script:

tts4

To change the voice while a call is in progress and the script is being heard by the recipient, you would use the <voice name=”David”>This is David.</voice> command.  For example, if the first portion of the call script needed to be spoken by a US English female voice, but the second portion needed to be spoken by a Spanish male voice, you would use something similar to this within your call script:

tts5

4. Test the Call Script

Once you have your call script adjusted and would like to place a few test calls to ensure the script flows fluently, you would replace your TTS parameter value of: Hello, this is a test. with:

Using the TextToSay Speech Control Commands within your call script helps create a less robotic sounding voice in addition to reducing any confusion as to what was played to recipient during the call.  Adjusting the script also helps reduce the need for an individual to record the script and upload it to your license key.  If you have multiple scripts, frequently changing scripts, or specific language needs such as Spanish, French, or German using the TextToSay Speech Control Commands make creating a custom TextToSay call script easy and quick.

For more information regarding our Phone Notify! API such as methods, programming examples, and available commands to use during the call, please visit: http://wiki.cdyne.com/index.php/Phone_Notify!

Leave a reply