With the exception of adding (various) LED indicator lights and high fidelity stereo playback, USB soundcards remained unchanged over the past decade. In the past, if you had a Turtle Beach or Sound Blaster PCI soundcard, it was the same quality as a USB soundcard. With the introduction of the innovative SpeechMatic MultiAdapter, that is no longer the case. We can finally say that one USB soundcard is better than any other PCI or USB soundcard. The SpeechMatic MultiAdapter literally broke new ground by introducing several new technologies. SpeechWare prefers not to refer to the SpeechMatic MultiAdapter as a USB soundcard because it is so much more. Besides the usual USB soundcard amenities, the MultiAdapter also includes a number of unique and exclusive features including:
• Auto gain control
• Auto EQ
• Dedicated circuit board with NaturallySpeaking algorithms
• De-clicking algorithm
• Detachable USB extension cord
• Broadband noise filtering
• Bypass mode
When you run the NaturallySpeaking Audio Check, you can see the blue bar bouncing up and down even as you do your best to maintain a steady volume level. Simply put, maintaining a steady volume level is not possible. Because of this, NaturallySpeaking measures your highest and lowest volume levels and averages the two scores for a less-than-perfect estimate. The SpeechMatic MultiAdapter auto gain control is similar to the auto volume control found on cassette recorders as far back as the early 70s. It is extremely fast and effective. To give you some idea of how effective it is, when using the MultiAdapter with most headset microphones, you can actually remove the headset, place it on your desktop and continue dictating from about 12 inches away. Your accuracy won’t be quite as high as wearing the headset, but trying that with any other soundcard would likely result in unacceptable accuracy. Because of built-in amplification, the SpeechMatic MultiAdapter auto gain control is ideally suited to soft voices. It can double or even triple the pickup range of short range microphones such as the Sennheiser MD431 II.
As you move away from a microphone, the volume level of your voice doesn’t just drop. Your voice also begins to sound thin, or less robust (for lack of a better term). Auto EQ boosts your vocal midrange as you lower your voice or move away from the microphone. In other words, simply amplifying your voice as you speak lower isn’t quite enough. Auto EQ and auto gain control work hand-in-hand.
With other soundcards, leaving your microphone on while clearing your throat or setting a drink down on the table can result in unwanted dictation because NaturallySpeaking interprets those sounds as words. NaturallySpeaking 11 introduced a "Nothing But Words” algorithm that works fairly well at differentiating dictation from noise. Unfortunately, it can take up to 4 seconds to process that information. The MultiAdapter de-clicking algorithm works in much the same way, but only takes about 1/10 of a second and can prevent unwanted noise from even reaching NaturallySpeaking.
Unlike other USB soundcards, the SpeechMatic MultiAdapter actually changes the characteristics of your voice. If you play back a recording of your voice in the greenlight mode, with all the specialty effects, it will sound a bit thin because the sound of your voice is being altered (tweaked) to increase your NaturallySpeaking accuracy. While this is effective for speech recognition, it is not ideally suited to uses such as VoIP and gaming. For this reason, the MultiAdapter additionally includes a blue light bypass mode. This, as the name implies, bypasses the special effects. Note that for the purposes of speech recognition, most of our tests were performed exclusively in the greenlight mode which is also perfect for end-users with soft voices and can even extend the pickup distance on short range microphones like the Sennheiser MD431 II.
Bottom line:
The SpeechMatic MultiAdapter is significantly higher priced than other USB soundcards because it is unlike any other USB soundcard and not just a USB soundcard. You might not be able to justify the expense for consumer use but if speech recognition plays a significant role in your workflow and you’re not already using a SpeechWare TableMike or TravelMike (which includes this technology), this is a must-have soundcard. While each end-user will experience different results, our tests, showed accuracy improved by a full 1% over other USB soundcards. That may not sound like much, but it’s the difference between 97% and 98% accuracy (a noticeable improvement). These results actually meant more than 25% fewer errors for our testers!