The ultimate Android phone audio benchmarking thread

[Jerry Hildenbrand]

My soundcard test results. All you need is for it to be better than the phone you're testing​

You can benchmark the audio hardware in your Android just like you can benchmark any other hardware, but the method is a little more complicated and fun.

Also remember, benchmarks never tell the whole story, and things companies can do to make the audio output sound better can have an adverse effect on the numbers. Just like an SoC, tuning audio hardware is pretty damn important.

Anyhoo, if you want to get started benchmarking your phone the same way I (and almost every other blog that benchmarks audio) here's how to get started.

Wait. I lied. One last note — the best way to benchmark audio equipment is the same as the best way to benchmark any equipment — with dedicated hardware instead of software. You can buy audio benchmarking equipment with a starting price of about $7,500 USD. It goes up from there. This is why most people use software and/or cheaper audio input hardware.

What you'll need:

• A computer running Windows. It can be your Windows desktop/laptop, a Mac with Windows loaded on it, or a VM done the right way with access to the physical hardware (including the USB subsystem if you buy a separate audio input device)
• RMAA software. You can get that here RightMark Audio Analyzer. Products. Audio Rightmark it's free.
• A good piece of recording software. I use Audacity. You need to visually monitor the levels so you can adjust input level and make sure things aren't clipping. Audacity is free.
• Your Android phone
• A sound card that's not totally crummy.
• A cable to send the output from your phone to the input of the sound card.

Your sound card

You might be able to use the sound card built into your motherboard, but chances are unless the box it came in talked about how excellent the audio is on it, it's going to be pretty noisy. Any noise your sound card injects into the system is going to affect the readings from your phone. Most cheap on board sound cards introduce way too much noise and crosstalk into the sound, and wreck your results if the phone you're testing has quality audio hardware inside it.

You can bypass all this by buying a USB stand alone audio input device with it's own sound card. The more you spend, the better chance you'll get something good. If you want something really great, buy a standalone DAC and amp for your PC. If you were to ask me which product I would recommend for doing these tests, I would point you at this Amazon.com: Focusrite Scarlett 2i2 USB Recording Audio Interface: Musical Instruments

Of course, if your MoBo has a high quality DSP/amp (creative has some nice ones) you're good.

Doing the tests

RMAA creates two files when you want to benchmark audio. (Be sure to RTFM for RMAA, because there's a lot of stuff in the software and you'll need to know what to look at and what buttons to push.) You'll have a calibration file and a signal file. Copy both of these to your phone.

Open your phone's audio player, and play the calibration wav file. It's a horrible noise, and it's what you'll use to adjust your input gain.

Connect your phone audio output (though the headphone jack) to the input on your audio card. Make sure you select the correct input in your recording software. Start recording, then play the calibration file on your phone. Let it finish, wait a couple seconds, then stop recording. Did the volume clip (greater than 0 dB gain)? If so, do it again after you turn down the level on your input. Fiddle with it until the calibration wav file is as close to max without clipping. If you use Audacity, yuo can see this up at the top. If the meter goes red, you're clipping. If it stays green, you're not. If it's yellow, you're close to the max. Shoot for yellow.

Once you have the input level calibrated, delete the recordings but DON'T TOUCH THE LEVELS IN YOUR RECORDING SOFTWARE or you'll have to start all over again.

With your phone still connected, queue up the test file in your music player on the phone. Start recording in your software, then begin playback from the phone. Let the whole thing record, then wait a couple seconds before stopping the recording. Save this recorded file, because that's what you're going to be analyzing in RMAA.

Back to RMAA, hit the button to analyze a file. Pick the recording you just made, and let it run. When it's finished, you'll get a pop up telling you to pick a slot. Pick slot one and hit OK. Boom. There are your numbers.

You can (and should) benchmark your sound card, too. With an internal sound card, just playback the created test file in RMAA to see the numbers. For an external sound card, you'll do the same thing you did with your phone. Send the output from your sound card or audio interface back to its own input with a loopback cable. Calibrate it with the calibration wav file the same way, and record the test file playback the same way you did with the phone. That's the file you'll test in RMAA.

About cables

If you use crappy cables (especially when connecting an external audio interface) you're gonna get crappy results. Be sure to get something properly built and shielded, then make sure they aren't behind your desk touching any power or network cables.

If your audio input uses AC power (and this includes an on board sound card inside your computer) be sure to have a filter on the AC side.

Sound is electricity. Electricity is magnetism. You place two magnets close to each other, and weird sh*t happens. Same thing with improperly shielded cabling.

Comparing results

I just finished testing the HTC 10 — here are the results and the parameters I use for a phone test:

​

If you test the HTC 10 on your equipment the numbers WILL be different.

All we can really do without buying the multi-thousand dollar testing equipment is use the same audio input for everything we test, then evaluate it with headphones. That's important. Your benchmarks mean nothing when compared to mine, unless you are using the exact same hardware and running the exact same software in the exact same room attached to the exact same plug in the exact same wall. RMAA grades your test results (poor/good/excellent) based on the file you play in it through the hardware you're using to play it. Use a good sound card and you can generally trust the grade you get.

 

[Kevin OQuinn]

I have "followed" this thread.

 

[Kevin OQuinn]

Are you recommending the FocusRite Scarlett external board over the Samson you use because of performance or simplicity or both?

They are both priced relatively the same it seems. Assuming the performance is the same, are there things the Samson will do that the Scarlett will not (like simultaneous line in with separate left/right XLR balanced outputs)?

EDIT: Answered my own question with research. The FocusRite will do exactly what I asked, but uses TRS for the balances outputs.

 

[Jerry Hildenbrand]

The Samson has two things against it:

1. Quality is hit or miss. I'm on my second one. The first one went back because touching any XLR cable used for input one made an audible input buzz through every output. This one has a wonky pan adjustment pot on input 4, so I set it once and just don't touch it anymore, and only use it when I have to use it. I miss my Yamaha mixer, but I got drunk and had it out in the rain so it's my own damn fault. One day I'll spring for another.

2. Simplicity. The Samson has 7 different adjustments for gain and compression on 4 input channels, as well as several different ways to mix the output signal. This is handy if you're trying to mix a guitar track into something already recorded (why I use it) or mixing a jam session on your porch, but it's a PITA if you don't need to do any of that.

The Focusrite just works, and is really easy to adjust to your liking. It's also easy to adjust when trying to get at the edge of clipping on the input side. The Samson (or any mixer) isn't.

Really, any good interface will work. Chances are anything but the very cheapest you can buy will be better than the phone you're testing as long as you've used good cables and hooked everything up without making a rats nest of cables that touch each other.