ClearVoice(TM) & ClearMusic(TM) technologies for voice/music enhancement in noisy environments

ClearMusic^TM & ClearVoice^TM Technologies

Engineers and digital signal processing experts in one of our co-founding companies, Refined Audiometrics Laboratory, LLC, have spent the last 10 years investigating the physics of human hearing in excruciating detail. This research has culminated to a very robust mathematical understanding and extremely sophisticated modeling of psychoacoustics.

We call our technology Spectral Importance Processing^TM, and it will be obvious as to why, if you are willing to read a bit more about it here below.

Our advanced proprietary mathematical models are the basis for the development of our advanced signal-processing algorithms that comprise the platform of our ClearMusic^TM music enhancement technology, as well as of its voice-related derivative, ClearVoice^TM. An added outgrowth of this research is our ability to dynamically modify the spectral loudness of sounds, in an ear-safe manner, so that one can clearly communicate even in the presence of loud ambient noise conditions as you would find on busy streets, intersections, office lobbies, train stations, etc. Voice and music have very different spectral characteristics and are perceived differently in the presence of noise.

Most communications systems utilize some form of noise cancellation at the talker’s end, to avoid passing his or her local noise over the air to the listener. But typical systems out there do not take account of the listener’s ambient noise environment. This is where ClearMusic^TM and ClearVoice^TM technologies come into play for music and voice, respectively. And the difference is immediately striking. Simply raising the volume level in the listener’s ear is insufficient and not ear-safe. Ear safety means paying attention to the overall signal level to avoid presenting over-loud signals to the ear. Over-loud signals can threaten the fragile hair cells in the cochlea leading to eventual hearing loss. Being ear-safe means that we never present over-loud signals to the ear.

Similarly, using selective equalization, something that other companies do, is really nothing more than raising the volume level in selected frequency bands –which is no better than raising the volume. Sure, you might be able to overcome some ambient noise in this manner. But when the speaker presents loud sounds, those too will become over-loud and potentially damaging to hearing. And then, different noise environments need radically different equalization settings. This is clearly not a good solution.

What is needed instead is a "psychoacoustic compression", so that only the faintest sounds get amplified, while the louder sounds are passed unchanged. And this must occur in a manner as to counteract the spectral components of ambient noise, which varies widely between e.g. street corners and train stations.

The graph here above shows what we mean by "psychoacoustic compression". The human brain is stimulated by signals, which are received differently by both ears, and combines the resulting stimuli in a very complex way so as to produce perception of locality, intensity, loudness, pitch, timber, spatial orientation, depth, etc.

As you can clearly see on the graph, the amount of compressive gain varies in a non-linear fashion as the incoming loudness grows. The graph also illustrates that at the loudest incoming levels one does not need much gain (because that's the nature of audio compression), while the lower loudness levels need increasing amounts of gain. Anything to the left of where the green curve touches the X-axis is below the threshold of hearing, and would normally be completely inaudible. One can hear examples of that during the scenario recording of the train here below, where it seems the voice is not even present at times. It is there, but it is below the elevated threshold caused by the noise.

Voice-related examples:

Here below are four links to distinct WAV files recorded in two different scenarios. These are examples of the same speech played back (or received over a handset) in the presence of serious ambient noise at the receiver end - one scenario is with a passing train and the other one is with the presence of a loud vacuum cleaner. In both cases, one file is recorded without, and the other is recorded with ClearVoice™ correction technology. We strongly recommend you listen to all of them with a set of stereo headphones instead of just using the speakers of your computer. The reason why these examples work best when played back in stereo headphones is that the left ear in this case is presented with the speech audio, while the noise is presented to both ears, slightly attenuated in the left ear, to simulate the shielding effects of actually holding a cell-phone handset to the left ear.

You will notice immediately that the recordings made with ClearVoice^TM correction technology continue to be intelligible, even at the maximum noise levels. On the other hand, the recordings made without ClearVoice™ corrections show the voice fading hopelessly into the background. No manual adjustments to speech volume were made in either case. All other parameters are scrupulously kept identical in all these recordings so that one can compare "apples with apples" and "oranges with oranges".

Music-related examples:

The following recording shows the impact that ClearMusic^TM technology can have on the enjoyment of music played under noisy ambient conditions. The recording is comprised of two 30-seconds segments.

In the recording, a certain piece of music is first played over 30 seconds while suddenly a train is heard passing by and the train's noise has been recorded in stereo in order to simulate the psychoacoustic impression of spatial information received by the brain. It is clear that at the peak of the train noise the music "disappears" and gets "drowned" in the excessive ambient noise. The same exact piece of music and the very same exact train noise are identically included in the 2nd 30-seconds segment of the recording.

This time, however, ClearMusic^TM corrections are imposed in real time upon the aggregate audio output of the system. One can still hear the train, of course, however in the second segment the listener continues to hear and enjoy the music, no matter what. As in the voice-related recordings earlier in this page, please use your stereo headset to listen to this recording.

Notice that the overall noise level has not diminished in the "before" and "after" recordings. The noise is as present as it ever was. Rather, our Spectral Importance Processing^TM algorithms elevate the perceived importance of the voice and music to the brain, in a spectrally sensitive and ear-safe manner, to allow your brain to perform psychoacoustic noise corrections.

And, by the way, for the reader who may be wondering what would happen if traditional noise cancellation technology was used instead, the answer is simple but it may be surprising. Even if one could have perfect noise cancellation in the single earpiece, the ambient noise entering the opposite ear still obscures speech and music in your brain. Remember that with a handset on one ear, you are not configured as when wearing an expensive, cumbersome, dual-earpiece headset with traditional noise cancellation. Consequently, only Spectral Importance Processing^TM technology can overcome ambient noise for handsets and portable music players in this way.

We believe that the recordings speak for themselves.

Wouldn't you want your next handset or music player to have this technology embedded in it?

ClearMusic^TM and ClearVoice^TMtechnologies can have a huge positive impact on the usability of an audio system under certain circumstances. These technologies are respectively very useful in playback of music on portable players (e.g. iPods, iPhones, smart phones, etc.), as well as in the operation of cellular handsets, cordless phones, land-mobile radios, TV sets, radio sets, earpieces, and even headsets like the ones used by military forces, or SWAT teams, or even teams of first responders in cases of emergency. In general, ClearVoice^TM and ClearMusic^TM technologies are not just for voice applications; they can make a huge difference for the user when they are embedded in any piece of audio equipment that is bound to be used in noisy environments.

ClearMusic^TM technology and its voice-related derivative, ClearVoice^TM, comprise sets of advanced mathematical algorithms, and optimized handcrafted embedded DSP (digital signal processing) software that runs adaptively in real time at the output of the system's vocoder or audio path. Despite its powerful functionality, the real-time software for both technologies can be implemented & easily integrated inside the available computational and memory resources of a typical handset, smartphone, or music playback device. Technology licenses are available in embedded platforms that enhance either voice, or music, or both.

Please contact SpectroDynamics today to find out how to embed this advanced DSP technology into your products. It will bring an extraordinary competitive advantage to your products, because your customers will be able to enjoy using them even in very noisy ambient circumstances.