Wiki » History » Version 2
Chris Cannam, 2014-07-18 07:57 PM
| 1 | 1 | Chris Cannam | h1. Wiki |
|---|---|---|---|
| 2 | 1 | Chris Cannam | |
| 3 | 1 | Chris Cannam | Trying this out in the "Piano Evaluation of the Silvet Note Transcription plugin":/projects/silvet/wiki/Piano_Evaluation_for_Level_Normalisation |
| 4 | 1 | Chris Cannam | |
| 5 | 1 | Chris Cannam | h3. First attempt |
| 6 | 1 | Chris Cannam | |
| 7 | 1 | Chris Cannam | As of commit:e36fe9312ad4 |
| 8 | 1 | Chris Cannam | |
| 9 | 2 | Chris Cannam | The aim is just to make the RMS level across a few seconds of audio tend toward some target. |
| 10 | 2 | Chris Cannam | |
| 11 | 2 | Chris Cannam | We have a target RMS value T (example 0.05). Start with an initial gain G equal to 1. |
| 12 | 1 | Chris Cannam | |
| 13 | 1 | Chris Cannam | At each sample: |
| 14 | 1 | Chris Cannam | |
| 15 | 1 | Chris Cannam | * Update the RMS level of the past 4 seconds of audio |
| 16 | 1 | Chris Cannam | * Find the gain G' that would be necessary to make that level equal to T (i.e. T / rms) |
| 17 | 1 | Chris Cannam | * Update our stored gain G to move it 1/N of the distance from G to G' (where N is 0.5 seconds in sample count) |
| 18 | 1 | Chris Cannam | * Return the sample scaled by G |