Machine Learning for the Inverse Control of FM Synthesis ROSA GARZA - - PowerPoint PPT Presentation

machine learning for the inverse control of fm synthesis
SMART_READER_LITE
LIVE PREVIEW

Machine Learning for the Inverse Control of FM Synthesis ROSA GARZA - - PowerPoint PPT Presentation

Machine Learning for the Inverse Control of FM Synthesis ROSA GARZA MENTORS: DR. EDGAR BERDAHL & ANDREW PFALZ CCT REU 2017 LSU CENTER FOR COMPUTATION & TECHNOLOGY (CCT) Frequency Modulation (FM) Synthesis Invented by John Chowning


slide-1
SLIDE 1

Machine Learning for the Inverse Control of FM Synthesis

ROSA GARZA MENTORS: DR. EDGAR BERDAHL & ANDREW PFALZ CCT REU 2017 LSU CENTER FOR COMPUTATION & TECHNOLOGY (CCT)

slide-2
SLIDE 2

Frequency Modulation (FM) Synthesis

  • Invented by John Chowning in 1967

*t is an array of +me Control Signals:

  • Carrier Frequency (CF)
  • Depth (D)
  • ModulaEon Frequency (MF)
slide-3
SLIDE 3

Current Sound Design Procedure

AddiEve / FM Synthesizer Synthesized Sound Control Signals

slide-4
SLIDE 4

Research Goal

Inverse Control of FM Synthesis Synthesized Sound Target Sound

slide-5
SLIDE 5
slide-6
SLIDE 6

Long Short Term Memory (LSTM) Recurrent Neural Network (RNN)

Hyperparameters:

  • Learning Rate (1e-4 – 1e-7)
  • Number of Unrollings (1,5,10,15,50,100)
  • Epochs (1,5,20,100)
  • Number of LSTM Layers

CalculaEng Loss: Mean Squared Error

  • Goal: Low loss (close to 0)
slide-7
SLIDE 7

First Test

1 Control Signal Audio LSTM Loss Label

slide-8
SLIDE 8

Second Test

3 Control Signals Audio LSTM Loss Label Fully Connected Layer

slide-9
SLIDE 9

Third Test

LSTM Audio Audio LSTM Loss Fully Connected Layer FM Synthesizer

slide-10
SLIDE 10

Small Edit to Third Test

Fourier Transform

  • f LSTM

Audio Audio LSTM Loss Fully Connected Layer FM Synthesizer Fourier Transform

  • f Audio
slide-11
SLIDE 11

Fourth Test

LSTM Audio Audio LSTM Loss Fully Connected Layer FM Synthesizer

slide-12
SLIDE 12

Most Recent Data

Seeing a loss of 0.002

slide-13
SLIDE 13

Piano Input Data

slide-14
SLIDE 14

Acknowledgements

Thank you to for this opportunity to be a part of the Center for ComputaEon & Technology (CCT) at Louisiana State University (LSU) REU 2017. Thank you to my graduate student, Andrew Pfalz, Dr. Berdahl, and Dr. Jesse Allison. Also to my family and friends for their support throughout my summer research experience. This material is based upon work supported by the NaEonal Science FoundaEon under award OCI-1560410 with addiEonal support from the CCT at LSU.

slide-15
SLIDE 15

Ques%ons?