You can now read part of an audio file with the new samples option for Read.

$\mathrm{audFile} ≔ \mathrm{FileTools}:-\mathrm{JoinPath}\left(\left[\mathrm{kernelopts}\left(\mathrm{datadir}\right)\, ''audio''\, ''ViolinThreePosVibrato.wav''\right]\right)\:$

$\mathrm{violinNote\_sub} ≔ \mathrm{Read}\left(\mathrm{audFile}\, \mathrm{samples} \= {10}^3..60\cdot {10}^3\right)$

You can create many audio effects with convolution. For example, you can add reverb, or simulate how audio would sound in a specific room with its convolution with an impulse response.

For Maple 2020, AudioTools:-Convolution is significantly faster (for the default case) for those CPUs where the Intel IPP libraries can run optimized code.

Consider this violin note (to interact with the audio playback buttons below, open this page as a worksheet)

$\mathrm{violinNote\_full} ≔ \mathrm{Read}\left(\mathrm{audFile}\right)\left[..\,1\right]\:$

$\mathrm{Preview}\left(\mathrm{violinNote\_full}\,\mathrm{output} \= \mathrm{embed}\right)$

Read the kernel for the convolution—a "ding" sound.

$\mathrm{kernel} ≔ \mathrm{Read}\left(\mathrm{FileTools}:-\mathrm{JoinPath}\left(\left[\mathrm{kernelopts}\left(\mathrm{datadir}\right)\, ''audio''\, ''ding.wav''\right]\right)\right)\:$

$\mathrm{Preview}\left(\mathrm{kernel}\, \mathrm{output} \= \mathrm{embed}\right)$

Convolution of the violin note with the ding sound (on an i7-8650U processor, Maple 2019 takes 2.6 s, while Maple 2020 takes 0.02 s)

$$\begin{gathered} \mathrm{tt} ≔ \mathrm{time}\left[\mathrm{real}\right]\left(\right)\: \\ \mathrm{violinNote\_conv} ≔ \mathrm{Convolution}\left( \mathrm{violinNote\_full}\, \mathrm{kernel}\right)\: \\ \mathrm{time}\left[\mathrm{real}\right]\left(\right)-\mathrm{tt}\; \end{gathered}$$

$\mathrm{Preview}\left(\mathrm{violinNote\_conv}\,\mathrm{output}\=\mathrm{embed}\right)$

Moreover, AudioTools:-Convolution now supports multi-channel audio. For audio with M channels, the kernel can be specified as an M-column Matrix or Array. Each channel of the kernel is applied to the corresponding channel of the audio.

You can now create 32-bit and 64-bit WAV files.

$\mathrm{aud\_float} ≔ \mathrm{Create}\left(\mathrm{rate} \= 11025\, \mathrm{duration} \= 5\,\mathrm{float}\=\mathrm{true}\,\mathrm{bits}\=32\right)\;$

$\#\mathrm{Write}\left(''aud\_float.wav''\,\mathrm{aud\_float}\right)$

You can now write audio files with sample rates of up to 2³² - 1 Hz = 4.29 GHz.

$\mathrm{aud\_high\_sample\_rate} ≔ \mathrm{Create}\left( \mathrm{rate} \= 2^{21}\right)$

$\#\mathrm{Write}\left(''aud\_high\_sample\_rate.wav''\,\mathrm{aud\_high\_sample\_rate}\right)$

You can now quickly generate audio filled with white noise.

$\mathrm{aud\_noise} ≔ \mathrm{Create}\left(\mathrm{noise} \= \mathrm{true}\, \mathrm{duration} \= 1\right)\;$

$\mathrm{Preview}\left(\mathrm{aud\_noise}\,\mathrm{output}=\mathrm{embed}\right)$

PCM audio is asymmetric—the highest possible value is not equal in magnitude to the lowest possible value. For example, 16-bit audio ranges in value from -32768 to +32767. This means that positive half-cycles of a waveform are compressed compared to negative ones.

AudioTools:-Write and AudioTools:-Read now offer new options to change the mapping of minimum and maximum values in the internal representation of an audio file to external values when written to a PCM WAV file. This now reflects the reality of working with PCM audio.

$$\begin{gathered} \mathrm{data} ≔ \mathrm{LinearAlgebra}:-\mathrm{RandomVector}\left(1000\, \mathrm{generator} \= 0..1\right)\: \\ \mathbf{for} i \mathbf{from} 1 \mathbf{to} 1000 \mathbf{do} \mathbf{if} \mathrm{data}\left[i\right] \= 0 \mathbf{then} \mathrm{data}\left[i\right]≔-1 \mathbf{end} \mathbf{if} \mathbf{end} \mathbf{do}\: \\ \mathrm{aud\_sym} ≔ \mathrm{Create}\left(\mathrm{data}\right)\; \end{gathered}$$

$\min \left(\mathrm{aud\_sym}\right)\;\max \left(\mathrm{aud\_sym}\right)$

$\mathrm{aud\_path}≔\mathrm{FileTools}:-\mathrm{JoinPath}\left(\left[\mathrm{FileTools}:-\mathrm{TemporaryDirectory}\left(\right)\,''foo.wav''\right]\right)$

$$\begin{gathered} \mathrm{aud\_asym}≔\mathrm{Read}\left(\mathrm{aud\_path}\, \mathrm{mapping} \= \min \right)\: \\ \min \left(\mathrm{aud\_asym}\right)\;\max \left(\mathrm{aud\_asym}\right) \end{gathered}$$