Analyze Speech Formants

Start with an audio object:

In[1]:=

speech=ExampleData[{"Audio","MaleVoice"},"Audio"]

Out[1]=

00:00

00:02

Data in Notebook

Compute the max of the signal on each partition and display it on top of the waveform:

In[2]:=

ShowAudioPlotspeech,

opts

,ListLinePlotAudioLocalMeasurements[speech,"Max"],

opts



Out[2]=

Compute the formants of the signal, which represent the distribution of frequencies filtered by the vocal tract, on each partition and display them on top of the spectrogram:

In[3]:=

ShowSpectrogramspeech,

opts

,ListLinePlot[AudioLocalMeasurements[speech,"Formants"]]

Out[3]=

Simulate how the speech might change in a railway tunnel by creating a reverberation effect with an impulse response:

In[4]:=

model=ExampleData[{"Audio","IRRailwayTunnel"},"Audio"]

Out[4]=

00:00

00:05

Data in Notebook

In[5]:=

new=AudioTrim[AudioReverb[speech,model],Duration[speech]]

Out[5]=

00:00

00:02

Data in Notebook

The recording environment can boost or attenuate certain frequencies, thus altering the formants. Compute and visualize the new formants:

In[6]:=

ShowSpectrogramnew,

opts

,ListLinePlot[AudioLocalMeasurements[new,"Formants"]]

Out[6]=

Related Symbols

Publisher Information