Archive for the ‘MacSpeech Dictate’ Category
MacSpeech Dictate
(review date August 2008 by Kevin O’Gorman)
Introduction
This product, released in 2008, is speech recognition software for the Apple Macintosh. It allows creation of both simple and complex documents and control of most aspects of the computer just by using your voice. In conjunction with Apple’s VoiceOver (screen reader and screen magnification) technology this software provides strong support for students with physical and learning disabilities. Dictate uses the same speech technology as Dragon Naturally Speaking from Nuance and is customised for the Macintosh. It is compatible with Intel-based computers and Macintosh OS 10.5 Leopard. It takes advantage of the Macintosh operating system’s built-in support for speech recognition.
Dictate works by creating a speech profile of the user’s voice matched to a selected language. It allows creation of one or more profiles in different languages and language variants. Many languages are available. The version of the software for sale in Australia provides a number of English variants, including an Australian English option.
MacSpeech Dictate is sold as a boxed set of two DVDs with a quick start guide. There are also options to buy the software with a range of headsets including a wireless version. To ensure audio fidelity, the basic headset kit is provided with a USB audio adapter for Macintosh computers with only line-in inputs. To be recognised automatically, the adapter needs to be plugged in before starting up the computer. However, it is easy to use the Sound Panel in System Preferences to have the software recognize the headset afterwards. The basic headset was found to work well for the purpose.
Installation and Setup
Installation is a straight forward but lengthy process. The quick start guide is easy to read and contains sufficient information about installation. Additional documentation is available on the first DVD, including a 40 page user manual in PDF format. The first DVD is used to install the application and the second to install the language files. This second part can take up to an hour.
image of microphone setup dialog
Initially the user must calibrate the microphone and train the application to recognise their voice. Dictate provides wizards to help with calibration and training. Once this is done, the application remembers these settings from session to session. The microphone can be re-calibrated at any time. The first step in training is to create a speech profile. The user selects the language option to be used and enters a name for this profile. A prompt is then issued to read from a pre-prepared script in the selected language, which is to be read in the user’s natural reading style. This takes about ten minutes. Once the profile has been saved, the software can be used. Alternatively, another profile can be created. Even with this minimal training of the software, basic speech recognition was very accurate. Dictate can be configured in Application Preferences to either launch in the default profile or to allow selection from a list of created profiles.
If necessary, Dictate allows further training of profiles. This is done by reading additional scripts and by importing examples of the user’s writing style from Word and other text documents. The software is very sensitive if specialised vocabulary is used. Dictate provides a means for adding a vocabulary list of specialised terms using the built-in note pad.
image of voice training window
Using Dictate
Documentation is mostly electronic, with the quick start guide, user manual and Tips and tricks window available from the help menu. The quick start guide contains voice commands for commonly used punctuation, while the user manual contains a comprehensive list of all voice commands. Dictate also allows creation of customised voice commands for specific purposes.
Dictate works in a modal fashion and can be opened in Command or Dictation mode. Command mode allows control of system functions and applications, while Dictation mode allows creation and editing of textual documents. A set of global voice commands is available in either mode. A floating Available Commands window provides a list of common global and application specific commands. The list of available commands varies depending on the mode being used. Dictate also allows control of applications by their menu commands.
Dictate provides a simple note pad for capturing speech, but material can be dictated directly into a Word document, another word processor application or text editor. It was found to work very well with Apple’s default word processor TextEdit. The initial draft of this document was created using Dictate with very few recognition errors. It must be kept in mind that dictation is session based. That is, it is not possible to use Dictation mode to edit pre-existing text. However, it is possible to select, copy and paste text from a previous session in Command mode. The text for each session appears to be stored in some sort of temporary cache that allows the software to perform editing functions.
The level of accuracy can be adjusted in General Preferences. A range from faster and less accurate to slower and more accurate can be selected.
Dictate allows the microphone to be switched on and off from a floating control button or a menu command. However, the software can also be placed into a sleep mode with a voice command, allowing it to be “woken up” with the keyboard or mouse.
I found the Australian profile to be quite accurate in interpreting my accent. However, having been born in London, I found the UK English version to be more accurate.
The company has stated in release notes that spelling and correction functions will be features added in a free update late in 2008.
Issues
several problems were encountered with the software. in the first two weeks of use, there were several crashes-about one in every three sessions. This usually happened in Command mode when using different combinations of global commands. Dictate provides a built-in feedback mechanism for reporting application crashes and the company was quick to respond to feedback to try to identify the problem. Since installing an update to the program, this problem has significantly decreased.
Another problem occurred when trying to create new profiles in other language variants. It was not possible to create profiles in US and UK English without crashing the application. MacSpeech has provided a “Dictate data first aid” application on their website to address this problem that checks for corrupted files and downloads replacements. The download and information about its use is at the MacSpeech web site mentioned below. After running the data first aid process, it was possible to create multiple profiles in different language dialects, including UK English and US English.
A minor problem occurred when using a mouse, trackpad or keyboard to perform some functions while using Dictate. This sometimes caused the software to stop responding to my voice. It was then necessary to restart the application to continue dictating. MacSpeech does warn of this problem in the user manual.
Product Information
| Manufacturer | MacSpeech http://www.macspeech.com |
|---|---|
| Recommended Australian distributor | MacSense Australia Pty Ltd 34 Thomas Street Ashfield NSW 2131 Phone: (02) 9798-3288 Fax: (02) 9798-3668 Website: www.macsense.com.au/ |
| System requirements |
|
| Price | $320 for Macspeech dictate v1.0 and Plantronic headset and delivery |
