如何使用Java生成高质量的WAVE文件

Question

I am trying to generate a quality WAVE file using the Java Sound API.

First I used the following values with 8kHz sample rate. And the quality is 128kbps . But it is not generating quality audio. Some words are even not clearly pronounced.

static AudioFormat.Encoding defaultEncoding = AudioFormat.Encoding.PCM_SIGNED;
static float fDefaultSampleRate = 8000;
static int nDefaultSampleSizeInBits = 16;
static int nDefaultChannels = 1;
static int frameSize = 2;
static float frameRate = 8000;
static boolean bDefaultBigEndian = false;

AudioFormat defaultFormat = new AudioFormat(defaultEncoding, fDefaultSampleRate, nDefaultSampleSizeInBits, nDefaultChannels, frameSize, frameRate, bDefaultBigEndian);
AudioInputStream GeneratedAudio = marytts.generateAudio(text); //generate audio from text
AudioInputStream audio = AudioSystem.getAudioInputStream(defaultFormat, GeneratedAudio);
AudioSystem.write(audio, AudioFileFormat.Type.WAVE, new File("FileName.wav"));

So then I used the following values with 44.1kHz sample rate. Also showing the quality as 705kbps But the problem is when I generate audio with these values the generated audio is quality but there are some noises like cracking when playing the audio.

static AudioFormat.Encoding defaultEncoding = AudioFormat.Encoding.PCM_SIGNED;
static float fDefaultSampleRate = 44100;
static int nDefaultSampleSizeInBits = 16;
static int nDefaultChannels = 1;
static int frameSize = 2;
static float frameRate = 44100;
static boolean bDefaultBigEndian = false;

AudioFormat defaultFormat = new AudioFormat(defaultEncoding, fDefaultSampleRate, nDefaultSampleSizeInBits, nDefaultChannels, frameSize, frameRate, bDefaultBigEndian);
AudioInputStream GeneratedAudio = marytts.generateAudio(text); //generate audio from text
AudioInputStream audio = AudioSystem.getAudioInputStream(defaultFormat, GeneratedAudio);
AudioSystem.write(audio, AudioFileFormat.Type.WAVE, new File("FileName.wav"));

So what I need to know is how can I generate quality audio without having some cracking background noises using this Java sound API? I am very grateful for any help. Thanks in advance.

Answer 1

There are a number of things that could result in crackling: from over-driving the speakers, to overflowing the number of bits to overly large discontinuities in the data.

For 16-bit encoding (standard for "CD Quality audio"), all your PCM data points should lie within the range -32767 to 32767 (fits in a short ) prior to conversion of the PCM values to bytes.

If they do, then I would look next to make sure I had the byte order correct, and at the algorithm that is producing the PCM values. IDK what to recommend for testing that besides inspection. Maybe try generating a simple sine wave at various volumes and see if the PCM values are as expected, for starters?

16-bit, mono, at 44100 fps should sound pretty good. Any chance what you are describing as crackle might be a form of aliasing? (I don't usually think of that form of distortion as a crackle.)