# Developing DSP software, am I 10 years too late?

I'm interested in developing software related to digital signal processing. The most obvious ideas are audio and video editing. There's an overwhelming amount of audio and video editing software out there. I feel like developing such software would be in vain.

• What are the most lucrative frontiers in DSP software today?
• Do you think audio/video software still have room for improvement?
• there are plenty of effects for either audio or video signals that can still use improvement or just ideas of what things can be done. it's not just "editing" (like cut and splice and mixing). – robert bristow-johnson Nov 11 '16 at 3:56

DSP is a wide field and goes beyond audio/video processing. And, since new digital devices are developed in any field, DSP algorithm implementation, simulation and design is still a very important area.

Some examples that come to my mind beyond audio/video:

• wireless communications: for example there is plenty of research going on in 5G cellular (i.e. mobile phones) and 802.11 (i.e. WiFi) standards.
• medical devices: more and more devices operate (semi-)automatically, requiring sophisticated signal processing to interpret their measurements.
• DSL: as users search for faster internet, the research on DSL improvement is very important.
• Big Data and deep learning: On the one hand, it's about designing low-complexity algorithms to solve the problems of huge sizes. On the other hand, it's about providnig hardware that can do these operations quickly.

This list is not at all exhaustive, and there are many more fields, where DSP is absolutely necessary and new techniques are developed on a daily basis.

• digital radio. and communications like matched filter or $M$-ary stuff. like what if you have a message and sumbuddy breaks down the message into (or defines it as) a bunch of symbols and your transmitter farts out some unique signal for each symbol. now, in reception, you wanna choose which is the correct symbol based on the received signal. that's sorta cool. – robert bristow-johnson Nov 11 '16 at 3:53

Good illustrations of DSP in the earlier answers. Want to add a little more to your audio/video question. Although the audio processing field may appear saturated at a naieve glance, yet, to me, there is a huge renaissance in the field, when you consider, 'speech recognition' and 'speech processing'. DSP community has worked for more than 40 years on these problems. But it is now, that they are being integrateed in everyday products. This brings new challenges however. Imagine someone commanding a Siri to find something while working on an oil rig, or close to an airport. There are tons of noisy environments, where speech processing and recognition lags. We can imagine a future where robots respond to speech commands. Would need new signal processing paradigms and algorithms. Indeed, you can even start today, and instead of targetting an audio processing software, target a speech recognition engine. One with a limited dictionary lets say. That can recognize your custom commands better. I bet you can improve on the commercially available ones. Then try putting the commands in the dictionary that may be useful. For example commands for controlling IoT components. Link it to an IoT sensor/actuator. There.. You have a modern signal processing system, not available off the shelf, able to perform useful stuff, and can even be marketed if found robust enough.