These five common issues in signal chain designs in cost-sensitive audio solutions, if solved, will have everything working together as a balancing act.
There’s a growing demand for audio in almost everything these days as it has become the preferred interface for communication and education. With this comes the demand for high quality sound. When creating the optimal signal chain design, there are some major dos and don’ts that are often overlooked. Avoiding these mistakes means saving on cost. As the Director of Engineering for HARMAN Embedded Audio, I lead our division of HARMAN in embedded technology hardware, software, product design, and co-branding. We’ve worked with hundreds of companies to develop the perfect audio signature for their products, whether existing or new developments. With cost sensitivities more critical than ever during these times of economic turbulency, these solutions may be worth considering.
Mistake #1: Beginning without a targeted goal
One of the first mistakes that a lot of developers make in signal chain design is not having a targeted goal from the start. They may know they want audio for speakers or headphones, but audio can be tailored to achieve so many different results. Ask yourself if you are looking for crisp music, clear phone calls, or something else? The signal chain design needs to be specific to the intended use case. Having a target in mind is not only important for creating the best audio for each particular purpose, but it also drives cost savings. As there are infinite paths one can take to get to any stated goal, knowing your destination allows you to build the correct rudder to navigate along the best and most efficient path. The design goals might be the lowest cost that meets a specification, the highest performance in a given 3D space or a design for a harsh environment.
Mistake #2: Not determining what is important to the specific use case
While there are many considerations to make in signal chain design, identifying which features are most important for the intended use case is crucial. You will want to create the flow with the correct features for your chosen product or intended use. Features you may get from headphones will not be the same as features needed in a smart phone, a couch or stadium speakers. What is it about this particular project that you need to focus on when designing your signal chain? Consider the following for example:
- Overall frequency response: a flat or balanced frequency response is best in products like headphones, but in open spaces there will be a need for more bass
- Overall loudness: The system should achieve a loudness to provide visceral feel (stadium) or the proper loudness to enhance the intelligibility (phone or headset)
- Environment: The addition of environmental noises will need to be considered as that will mask certain portions of your intended audio delivery. Are there any regulations? Stadiums need to hit certain volumes, while headphones cannot exceed certain output levels for hearing related health reasons.
- Distortion: In most cases lower distortion is preferred as it can mask or change the perception of the audio. In some cases, distortion can be a benefit to the output in that most non-trained listeners perceive some distortion as overall loudness.
Mistake #3: Underestimating the hardware capabilities
Hardware can be a major cost-sensitivity; however, existing hardware does not completely limit the ability to add or improve audio. If there is no budget to replace hardware or no remaining space to build upon, we must strategically be able to make it work. Utilizing software is a great method to improve the sound quality in already completed hardware. We had several television manufacturers come to us looking to improve the audio of their products. We were able to take the product and hardware they already had and implement our internally developed software package that runs on the main processor in order to accomplish this without having to start from the beginning or design in new hardware.
This ability to add software is a powerful tool. In some cases, there is the assumption that software is an easy after-the-fact fix, but that isn’t always the case. If we start at the beginning, we can design and build software and hardware in tandem, thinking through capabilities and constraints as we go. When collaborating alongside Kohler to design their Moxie showerhead speaker, we had a blank slate to work from. This meant we were able to create a perfect blend of hardware and software that served as we needed to result in the best end product. Working together through this tends to be the most cost-effective option.
Mistake #4: Not ensuring that component capabilities match
Considering component capabilities and constraints is imperative. The signal chain is only as strong as its weakest link, so in design, you want to match the component capabilities. Once you have solved for this, you can look at specific tools that you can use moving forward toward your stated goals. For example, a smart amp can process internally. If an amp isn’t “smart,” you have to configure it separately. What components are you going to need to do this?
Knowing the proper tools and having them is important—and this can be driven by cost. Lay out a simple storyboard to show how the signal chain moves through the product. This is where it gets more complicated because most modern devices have several paths to work through, so they could have similar results but in the end the sounds aren’t the same. For example, if you’re switching from your Apple TV to your DVD player, you might get a different sound output. When you do these block diagrams, it is important you look at every single input and calculate gains to determine how much softer or louder you make each block. You will also need to proper algorithm in each block as these can change the signal level from one to the other.
As complicated as this all seems, you do not want to make it too complex. Processor MIPS and memory are so inexpensive, we tend to lean on them as a crutch. With so many tools being used, at some point it reaches a threshold; getting caught up using too many tools can also be an issue. Mitigate this by getting a good baseline measurement with no processing. At that point you can look at the toolbox to determine which algorithms are best for each block. Sometimes you will use the same tools, sometimes they will differ, but use a soft touch to ensure they are working together and not separately. Too many cooks in the kitchen are not only noisy but can ruin the meal.
Mistake #5: Forgetting to address gains in the system
Looking at the gains in the system as a whole, there needs to be consistency across signal chains. Blocks aren’t super devices that take any signal in and deliver any signal out—they’re limited electronically or mathematically. If there’s too much gain, it may be too noisy. If it is too high at any stage, there will be clipping in the system that will not allow the signal to get high enough. Work with one block at time. It is also crucial that you listen throughout the process and have others listening. As acoustic engineers, it is easy to miss things because of listening fatigue. Another’s ear may hear what you don’t.
Finding the right solution
These five common issues in signal chain designs in cost-sensitive audio solutions, if solved, will have everything working together as a balancing act. There are numerous ways that one can work to cut costs. For example, using HARMAN Embedded’s AudioEFX suite of post-processing algorithms allows for seamless blending of hardware and software. Each block can then be tuned to each specific device, each use case, and each environment. This tailored approach can address all above factors along the way, meaning avoiding any mistakes that will result in having to work backwards to adjust. Whichever path you choose to take, keep your ultimate goal in mind at each step of the way.
|Bruce Ryan is the Director of Engineering for the HARMAN Embedded Audio group based in Northridge, California and Shenzhen China. He has a Master of Science in Engineering and a Master of Business Administration. Bruce has worked at HARMAN for 20 years. His experience covers Acoustics, Electronics, Mechanical and Industrial design in automotive and consumer electronics. He currently leads the HARMAN design and development group dedicated to enabling high-performance, cost-effective audio capture and rendering.
For more Embedded, subscribe to Embedded’s weekly email newsletter.