What is the input data format of a sound card (in library `miniaudio` or others)?

Question

I am developing an acoustic simulation program. I can calculate the pressure wave (in Pa), but I cannot play it on my MacBook Pro. I'm using a helper lib miniaudio.

My questions are:

1. What is the input format of PC sound card? I mean, Is it correct to write a discretized sound pressure wave, which represented in physical unit $Pa$, into the buffer of sound card, and expect a proper behavior? (for general developement, or my current helper library selection miniaudio)

2. After search for quite a while, I still cannot find any useful resource for audio development. How can I find a development manual?

At this moment, my project works in cpp. Thanks!

Answer 1

1. Units of computer audio samples

Computer audio devices usually accept as input some form of linear pulse-code modulation. In such a format, to a first approximation, each sample represents a number in the range -1 to 1 that indicates the physical position of the cone of a loudspeaker within its travel range, generating audible noise as it moves back and forth.

The physical cone position correlates with the variations in air pressure known as "sound", since it is the source of the pressure variations, but computer audio formats do not directly deal with pressure in Pascals or whatever. You'll need to convert air pressure values to cone travel positions somehow, but doing so with high accuracy is a topic beyond the scope of Stack Overflow. (You could ask on Physics.SE.) A quick and dirty method is to just map whatever range of values you have to [-1,1] in the obvious way.

2. Generating audio with miniaudio

The miniaudio library has good documentation to get you started. The simple_playback_sine.c example demonstrates using the low-level API, in which individual audio samples are computed and passed to the audio device for playback. However, that sample also uses the wa_waveform API, making it just a bit more complicated than necessary.

So, I created a simplified example I called simple_playback_woowoo.c that computes the sample values directly. The key parts are:

1. Use ma_device_start and its prerequisites to arrange for a callback, data_callback, to be invoked every time the audio device is ready for another short clip of audio to play.

2. In data_callback, populate the pOutput array with sample values. I configured it to use the ma_format_f32 format, with two channels (stereo), so pOutput is an array of pairs of float sample values in [-1,1]. The frameCount parameter says how many frames (pairs of samples) are required.

3. I wrote a very simple sine wave generator that is the source of the samples. For fun, I also made the frequency vary over time using the same wave generator logic, so it makes a "woowoo" sound.

Here is the complete simple_playback_woowoo.c example:

/*
Demonstrates playback of a time-varying sine wave by directly specifying the sample values.

The intention is that a new miniaudio user who wants to learn the
low-level API can start here.  In contrast, `simple_playback_sine.c`
uses the `ma_waveform` API to generate samples, which makes it a little
bit harder to understand.

This example works with Emscripten.
*/

/* The encoding and decoding facilities are not needed here. */
#define MA_NO_DECODING
#define MA_NO_ENCODING

#define MINIAUDIO_IMPLEMENTATION
#include "../miniaudio.h"              /* ma_XXX, MA_XXX */

#include <math.h>                      /* floor */
#include <stdlib.h>                    /* strtod */

#ifdef __EMSCRIPTEN__
#include <emscripten.h>                /* emscripten_set_main_loop */

void main_loop__em()
{
}
#endif


/* Use a common format. */
#define DEVICE_FORMAT       ma_format_f32
#define DEVICE_CHANNELS     2
#define DEVICE_SAMPLE_RATE  48000


/* Describes a wave to produce, and where we are within that wave. */
typedef struct SineWave {
  /* Central value around which we oscillate. */
  double center;

  /* Amount to go above and below the center. */
  double amplitude;

  /* Frequency with which we repeat, in Hz. */
  double frequency;

  /* Current phase in [0,1). */
  double phase;
} SineWave;


/* Advance 'wave' by one audio sample of time, and return the value of
 * the wave form at that point. */
double nextWaveSample(SineWave *wave)
{
  /* Advance the phase. */
  wave->phase += wave->frequency / DEVICE_SAMPLE_RATE;
  if (wave->phase >= 1.0) {
    wave->phase -= floor(wave->phase);
  }

  /* Convert phase into amplitude. */
  return wave->center + ma_sind(MA_TAU_D * wave->phase) * wave->amplitude;
}


/* The output sound frequency will vary over time.  By default, it
 * runs from 250 to 350 Hz and back over the period of one second,
 * making a sort of "woo woo" sound, hence this file's name. */
SineWave frequencyWave = { 300, 50, 1 };


/* The output sound wave itself.  Its frequency varies. */
SineWave soundWave = { 0, 1 };


void data_callback(ma_device* pDevice, void* pOutput, const void* pInput, ma_uint32 frameCount)
{
    /* Each sample is represented by a single 'float' when using
     * 'ma_format_f32'. */
    float* pOutputF32 = (float*)pOutput;

    /* Frame counter in [0, frameCount-1]. */
    ma_uint32 curFrame;

    /* This callback is tied to the specific sample format and rate. */
    MA_ASSERT(pDevice->playback.format   == DEVICE_FORMAT);
    MA_ASSERT(pDevice->playback.channels == DEVICE_CHANNELS);
    MA_ASSERT(pDevice->sampleRate        == DEVICE_SAMPLE_RATE);

    /* Populate 'pOutput' with 'frameCount' frames. */
    for (curFrame = 0; curFrame < frameCount; ++curFrame) {
        /* Advance the frequency wave. */
        soundWave.frequency = nextWaveSample(&frequencyWave);

        /* Advance the sound wave to obtain the sound sample value
         * in [-1,1].  This represents, to a first approximation, the
         * physical location of the speaker cone within its range of
         * travel as it moves in order to generate audible sound. */
        float value = (float)nextWaveSample(&soundWave);

        /* Stereo output has two channels. */
        pOutputF32[curFrame * DEVICE_CHANNELS + 0] = value;
        pOutputF32[curFrame * DEVICE_CHANNELS + 1] = value;
    }

    (void)pInput;   /* Unused. */
}


int main(int argc, char** argv)
{
    ma_device_config deviceConfig;
    ma_device device;

    /* Command line can specify center, amplitude, and frequency. */
    if (argc >= 2) {
        frequencyWave.center = strtod(argv[1], NULL);
        if (argc >= 3) {
            frequencyWave.amplitude = strtod(argv[2], NULL);
            if (argc >= 4) {
                frequencyWave.frequency = strtod(argv[3], NULL);
            }
        }
    }

    deviceConfig = ma_device_config_init(ma_device_type_playback);
    deviceConfig.playback.format   = DEVICE_FORMAT;
    deviceConfig.playback.channels = DEVICE_CHANNELS;
    deviceConfig.sampleRate        = DEVICE_SAMPLE_RATE;
    deviceConfig.dataCallback      = data_callback;

    if (ma_device_init(NULL, &deviceConfig, &device) != MA_SUCCESS) {
        printf("Failed to open playback device.\n");
        return -4;
    }

    printf("Device Name: %s\n", device.playback.name);

    if (ma_device_start(&device) != MA_SUCCESS) {
        printf("Failed to start playback device.\n");
        ma_device_uninit(&device);
        return -5;
    }

#ifdef __EMSCRIPTEN__
    emscripten_set_main_loop(main_loop__em, 0, 1);
#else
    printf("Press Enter to quit...\n");
    getchar();
#endif

    ma_device_uninit(&device);

    return 0;
}