[DEV] Voodoo Sound for Transformer v1 for 3.0.1, 3.1. Feature requests and discussion
Asus EeePad Transformer hosts a very nice audio codec by Wolfson that powers both the headphone and speaker output.
Its the ultra low power WM8903
, implementing a midrange DAC (Wolfson-scale) that outputs a clean and musical signal, and my favorite Class W headphone amp, that gives a clean controlled and accurate output, and is able to drive full-size 300 Ohms headphones without distortion.
The transformer audio power stage is not as powerful as other device I tune so at maximum volume but I'm sure you'll be surprised by the capability of this amp if you compare it to any other mobile non-Voodoo device.
I might limit the maximum gain to +3dB to avoid saturation.
Voodoo sound is:
A control Android application
- Free version (semi broken on Honeycomb ATM, will fix soon): https://market.android.com/details?i...doo.controlapp
- Plus/Donate version: https://market.android.com/details?i...ntrolappdonate
version gives the ability to get full Voodoo
sound implementation without depending on a custom kernel. Root is required to load the additional driver.
version requires a kernel implementing Voodoo
sound GPL patches as you'll find plenty on XDA.
Features of the v1 :
I'll be very accurate and describe the exact change so far.
As usual, you get what you see.
What Voodoo sound driver v1 does on transformer is mapping volume to the headphone amplifier hardware control.
In the standard configuration, here is how it works: Most time only half or less of the Dynamic Range of the DAC is used (depending on the current output volume) by trimming the signal amplitude in a digital way.
What brings the driver v1 is a maximization of the digital volume, so the system works really on 16bit instead of 12 or even less at low volumes. And instead of that you control directly the analog gain of the Class W headphone amplifier built in the codec.
It allows to get more powerful sound, like to drive full-size high-impedance headphones at desired volumes, but for most people using sensitive headphones (same for in-ear) the sound is improved because there's far less hiss by reducing the analog gain.
When you reduce the noise level, the sound is more clear and defined.
I have some more leads in development for this codec, like:
- better oversampling quality (128x oversampling, require also clock reconfiguration on the SoC)
- alternative digital filter to reduce the very high frequency intensity ( > 20kHz) in order to reduce the listening fatigue.
WM8903 doesn't support digital effects like parametric EQ or 3D effect so I won't be able to activate that, but there's still some headroom to improve the quality and neutrality of the response, audiophile-style
Here are measurements I made without any modification, on 3.1:
- Better SRS presets for the speakers
- Maybe some volume boost for speakers if possible with reasonable quality.
- If usable and providing sufficient quality: customizable EQ & 3D effect using SRS software engine for the Headphone out.