Update README

README.md

@@ -1,23 +1,25 @@
 LASP
 ====
 
+![gif](https://i.imgur.com/avAppLA.gif)
+![gif](https://i.imgur.com/8KTOr2u.gif)
+
 **LASP** is a Unity plugin providing low-latency, high-performance and
 easy-to-use audio input functionality that is useful for creating audio
 reactive visuals.
 
-[Demo 1](http://radiumsoftware.tumblr.com/post/163009893309),
-[Demo 2](http://radiumsoftware.tumblr.com/post/163095570430),
-[Demo 3](http://radiumsoftware.tumblr.com/post/163137586857)
-
 Features
 --------
 
 - Low latency (less than 16 ms) audio input.
 - High performance audio analysis (peak level detection, RMS calculation) with
-  C++ native code.
+  C++ native implementation.
 - Three band (low, middle, high) filter bank useful for detecting rhythmic
   accents.
-- Supports Windows (WASAPI), macOS (Core Audio) and Linux (ALSA).
+- Dynamic normalization based on the most recent peak level.
+- Multi platform support via [PortAudio].
+
+[PortAudio]: http://www.portaudio.com
 
 System Requirements
 -------------------
@@ -34,30 +36,77 @@ to a project.
 
 [Releases]: https://github.com/keijiro/Lasp/releases
 
-How To Use
-----------
+You can also use [Git support on Package Manager] to import the package. Add
+the following line to the `dependencies` section in the package manifest file
+(`Packages/manifest.json`).
+
+```
+"jp.keijiro.lasp": "https://github.com/keijiro/Lasp.git#upm"
+```
+
+[Git support on Package Manager]:
+    https://forum.unity.com/threads/git-support-on-package-manager.573673/
+
+Audio Input Tracker Component
+-----------------------------
+
+The **Audio Input Tracker** component is used to receive audio input and
+control other components by normalized audio level.
+
+![GIF](https://i.imgur.com/ddCZSs5.gif)
+
+It tracks the most recent peak level and calculates the normalized level value
+based on the difference between the current level and the peak level. It only
+outputs an effective value when the current level is in its dynamic range,
+which is indicated by the gray band in the VU meter.
+
+### Filter Type
+
+Four types of filters are available: **Bypass**, **Low-Pass**, **Band-Pass**
+and **High-Pass**. These filters are useful to detect a specific type of
+rhythmic accents. For instance, the low-pass filter can be used to make a
+behavior that reacts to kick drums and basslines.
+
+### Dynamic Range (dB)
+
+This specifies the range of audio level normalization; The output value becomes
+zero when the input level is equal or lower than (*peak* - *dynamic range*).
 
-All the public methods of LASP are implemented in [`MasterInput`] as static
-class methods that can be called without any setup. Each of these methods has a
-[`FilterType`] argument and returns filtered results based on the argument (or
-just returns non-filtered results with `FilterType.Bypass`).
+### Peak Tracking
+
+When enabled, it automatically tracks the peak level as explained above. When
+disabled, the peak level is fixed at 0 dB. In this case, the effective range
+can be manually controlled by the **Gain** property, which is only shown when
+peak tracking is disabled.
+
+### Hold And Fall Down
+
+This adds a "peak-hold and fall down" behavior to the output value that is
+commonly used in VU meters. This is useful to make choppy animation smooth.
+
+![GIF](https://i.imgur.com/MEojdmD.gif)
+
+Scripting Interface
+-------------------
+
+LASP also provides audio input functionality via the [`MasterInput`] class. All
+the following methods are implemented as static methods and can be used without
+any setup.
 
 #### GetPeakLevel/GetPeakLevelDecibel
 
-`GetPeakLevel` returns the peak level of the audio signal during the last
-frame. `GetPeakLevelDecibel` returns the peak level in dBFS.
+`GetPeakLevel`/`GetPeakLevelDecibel` are used to get the peak level of the
+audio input during the last frame. `GetPeakLevel` returns the value in linear
+scale; `GetPeakLevelDecibel` uses dBFS instead.
 
-`AudioInput` automatically caches the results, so that these methods can be
+`MasterInput` automatically caches the results, so that these methods can be
 called multiple times without wasting CPU time.
 
 #### CalculateRMS/CalculateRMSDecibel
 
-`CalculateRMS` calculates and returns the RMS (root mean square) of the audio
-signal level during the last frame. `CalculateRMSDecibel` returns the RMS in
-dBFS.
-
-`AudioInput` automatically caches the results, so that these methods can be
-called multiple times without wasting CPU time.
+`CalculateRMS`/`CalculateRMSDecibel` are almost the same to
+`GetPeakLevel`/`GetPeakLevelDecibel` but use RMS (root mean square) instead
+of peak level.
 
 #### RetrieveWaveform
 
@@ -68,6 +117,18 @@ internal buffer. It's safe to use 1024 or less.
 [`MasterInput`]: Assets/Lasp/Runtime/MasterInput.cs#L16
 [`FilterType`]: Assets/Lasp/Runtime/MasterInput.cs#L9
 
+Tips
+----
+
+### Keep dynamic range as narrow as possible
+
+Although it's important to have a wide dynamic range for expressiveness, it
+tends to make animation slower and unclear. In general, it's recommended to
+keep the dynamic range as narrow as possible to make animation fast and
+accent-sensitive.
+
+![GIF](https://i.imgur.com/kEvGqEG.gif)
+
 Current Limitations
 -------------------