Microsoft Speech Platform

Text-to-Speech (TTS) Overview

ISpVoice Introduction

ISpVoice is the API for text-to-speech (TTS) in the Microsoft Speech Platform. Using this interface, applications can add the TTS support needed to speak text, modify speech characteristics, change voices, and respond to real-time events while speaking.

Applications obtain access to ISpVoice interface methods by creating a COM object. As the name implies, an ISpVoice object is simply a single instance of a specific TTS voice. Every ISpVoice object is an individual voice. Even if two different ISpVoice objects select the same voice token (for example "en-US_Helen"), each of the two voices can be changed and modified independently of the other.

Speak Text

When an application first creates an ISpVoice object, the object initializes to the default voice token that is specified at the following registry key: HKEY_CURRENT_USER\Software\Microsoft\Speech Server\v11.0\Voices\DefaultTokenId. This means that the new object is immediately ready to speak text, no special initialization is needed. At this point, applications can use Speak or SpeakStream to speak any Unicode text data.

Synchronous vs. Asynchronous Speaking

The two speaking functions can generate speech either synchronously (function does not return until text has completely spoken) or asynchronously (function returns immediately but continues speaking as a background process). Asynchronous operation is typically preferred if the application needs to do something else (highlight text, paint animation, monitor controls, or similar) while speaking. Otherwise, the simplest case is to speak synchronously.

Get Status Information

During asynchronous speech, applications can get current status information (text position, speech done state, bookmarks, etc.) in one of two ways. The simplest way is to periodically poll the ISpVoice object using the GetStatus method. The other way is to initialize the ISpVoice object so that it sends real-time events to the application as they happen.

Control Flow

As a convenience, most TTS applications allow users to temporarily suspend speech output. The Pause and Resume methods are typically called in response to a user initiated action.

Modify Voice Attributes

Often with TTS, voice output needs to be modified from its default setting. There are two ways to do this is; either by calling certain ISpVoice API methods, or by embedding XML that conforms to the Speech Synthesis Markup Language (SSML) Version 1.0 within the text to speak. Typically, the API functions are used as global settings that affect the speech independent of current selected voice or document that is spoken. While the SSML tags are usually used in much narrower scope, affecting only the spoken style in a single document. For information about modifying voice attributes using SSML, see Use SSML to Create Prompts and Control TTS.

Manage Audio Output

Audio output for TTS is not restricted to hardware sound card destinations. TTS functionality in the Speech Platform supports, either directly or indirectly, almost any audio configuration an application may require. Whether the destination is a PC sound card, buffer in memory, or a special telephony hardware, ISpVoice has several audio control methods to change the audio path from its default configuration.

ISpVoice Methods

Speak Text

Get Real-time Status

Control Flow

Change Voice Attributes

Process Events in Real Time (inherited from ISpEventSource)

Manage Audio Output

Miscellaneous