#1170 resolves issue with MP3 encoded audio raspy. Adds auto-clipping of PCM audio prior to encoding and adds 4x new 32-bit input sample rate selections to choose from, in addition to original 4x 16-bit input formats. (#1171)

Co-authored-by: Dennis Sheirer <dsheirer@github.com>
This commit is contained in:
Denny Sheirer
2022-03-16 05:00:07 -04:00
committed by GitHub
parent 90bc321aca
commit 7749fd10f1
23 changed files with 245 additions and 183 deletions

View File

@@ -24,14 +24,19 @@ import io.github.dsheirer.audio.AudioFormats;
import javax.sound.sampled.AudioFormat;
/**
* Enumeration of (input) audio sample rates for mono audio.
* Enumeration of MP3 input audio formats for mono audio.
*/
public enum AudioSampleRate
public enum InputAudioFormat
{
SR_8000(AudioFormats.PCM_SIGNED_8_KHZ_16BITS_MONO, "8000 Hz (no resample / default)"),
SR_16000(AudioFormats.PCM_SIGNED_16_KHZ_16BITS_MONO, "16000 Hz"),
SR_22050(AudioFormats.PCM_SIGNED_22050_HZ_16BITS_MONO, "22050 Hz"),
SR_44100(AudioFormats.PCM_SIGNED_44100_HZ_16BITS_MONO, "44100 Hz");
SR_8000(AudioFormats.PCM_SIGNED_8000_HZ_16_BIT_MONO, "16-Bit 8000 Hz (no resample)"),
SR_16000(AudioFormats.PCM_SIGNED_16000_HZ_16_BIT_MONO, "16-Bit 16000 Hz"),
SR_22050(AudioFormats.PCM_SIGNED_22050_HZ_16_BIT_MONO, "16-Bit 22050 Hz (default)"),
SR_44100(AudioFormats.PCM_SIGNED_44100_HZ_16_BIT_MONO, "16-Bit 44100 Hz"),
SR_32_8000(AudioFormats.PCM_SIGNED_8000_HZ_32_BIT_MONO, "32-Bit 8000 Hz (no resample)"),
SR_32_16000(AudioFormats.PCM_SIGNED_16000_HZ_32_BIT_MONO, "32-Bit 16000 Hz"),
SR_32_22050(AudioFormats.PCM_SIGNED_22050_HZ_32_BIT_MONO, "32-Bit 22050 Hz"),
SR_32_44100(AudioFormats.PCM_SIGNED_44100_HZ_32_BIT_MONO, "32-Bit 44100 Hz");
private AudioFormat mAudioFormat;
private String mLabel;
@@ -40,7 +45,7 @@ public enum AudioSampleRate
* Constructs an instance
* @param audioFormat for the specified sample rate
*/
AudioSampleRate(AudioFormat audioFormat, String label)
InputAudioFormat(AudioFormat audioFormat, String label)
{
mAudioFormat = audioFormat;
mLabel = label;
@@ -49,9 +54,9 @@ public enum AudioSampleRate
/**
* Default sample rate
*/
public static AudioSampleRate getDefault()
public static InputAudioFormat getDefault()
{
return SR_8000;
return SR_22050;
}
/**

View File

@@ -34,7 +34,7 @@ public class LameFactory
* @param setting to apply to the encoder
* @return lame encoder
*/
public static LameEncoder getLameEncoder(AudioSampleRate input, MP3Setting setting)
public static LameEncoder getLameEncoder(InputAudioFormat input, MP3Setting setting)
{
return switch(setting)
{
@@ -52,13 +52,13 @@ public class LameFactory
* @param sampleRate to resample to
* @return resampler.
*/
public static RealResampler getResampler(AudioSampleRate sampleRate)
public static RealResampler getResampler(InputAudioFormat sampleRate)
{
return switch(sampleRate)
{
case SR_16000 -> new RealResampler(8000, 16000, 4096, 512);
case SR_22050 -> new RealResampler(8000, 22050, 4096, 512);
case SR_44100 -> new RealResampler(8000, 44100, 4096, 512);
case SR_16000, SR_32_16000 -> new RealResampler(8000, 16000, 4096, 512);
case SR_22050, SR_32_22050 -> new RealResampler(8000, 22050, 4096, 512);
case SR_44100, SR_32_44100 -> new RealResampler(8000, 44100, 4096, 512);
default -> throw new IllegalArgumentException("Unrecognized sample rate for resampling: " + sampleRate);
};
}

View File

@@ -32,6 +32,9 @@ import java.util.Arrays;
import java.util.Collections;
import java.util.List;
/**
* Converts PCM audio packets to Mono, MP3 encoded audio
*/
public class MP3AudioConverter implements IAudioConverter
{
private final static Logger mLog = LoggerFactory.getLogger( MP3AudioConverter.class );
@@ -40,25 +43,33 @@ public class MP3AudioConverter implements IAudioConverter
private RealResampler mResampler;
private ByteArrayOutputStream mMP3EncodedFramesStream = new ByteArrayOutputStream();
private byte[] mOutputFramesBuffer;
private InputAudioFormat mInputAudioFormat;
/**
* Converts PCM 16-bit Little Endian audio packets to Mono, MP3 compressed audio.
* Constructs an instance.
*
* @param audioSampleRate for the desired input sample rate (resampled from default 8 kHz as needed)
* @param inputAudioFormat for the desired input sample rate and bit size (resampled from default 8 kHz as needed)
* @param setting to configure the LAME encoder
*/
public MP3AudioConverter(AudioSampleRate audioSampleRate, MP3Setting setting)
public MP3AudioConverter(InputAudioFormat inputAudioFormat, MP3Setting setting)
{
mEncoder = LameFactory.getLameEncoder(audioSampleRate, setting);
mInputAudioFormat = inputAudioFormat;
mEncoder = LameFactory.getLameEncoder(inputAudioFormat, setting);
if(audioSampleRate != AudioSampleRate.SR_8000)
//Resampling is only required if desired input sample rate is not system default of 8kHz
if(inputAudioFormat != InputAudioFormat.SR_8000 && inputAudioFormat != InputAudioFormat.SR_32_8000)
{
mResampler = LameFactory.getResampler(audioSampleRate);
mResampler = LameFactory.getResampler(inputAudioFormat);
}
mOutputFramesBuffer = new byte[mEncoder.getPCMBufferSize()];
}
/**
* Converts the list of PCM audio packets to MP3 encoded.
* @param audioPackets of PCM audio sampled at 8 kHz
* @return encoded MP3 audio
*/
public List<byte[]> convert(List<float[]> audioPackets)
{
List<byte[]> converted = new ArrayList<>();
@@ -70,7 +81,17 @@ public class MP3AudioConverter implements IAudioConverter
for(int x = 0; x < audioPackets.size(); x++)
{
byte[] bytesToEncode = ConversionUtils.convertToSigned16BitSamples(audioPackets.get(x)).array();
byte[] bytesToEncode = null;
if(mInputAudioFormat.getAudioFormat().getSampleSizeInBits() == 16)
{
bytesToEncode = ConversionUtils.convertToSigned16BitSamples(audioPackets.get(x)).array();
}
else
{
bytesToEncode = ConversionUtils.convertToSigned32BitSamples(audioPackets.get(x)).array();
}
int bytesToEncodePointer = 0;
int inputChunkSize = FastMath.min(mOutputFramesBuffer.length, bytesToEncode.length);

View File

@@ -58,7 +58,7 @@ public class MP3FrameInspector
public static void main(String[] args)
{
mLog.info("Starting ...");
MP3SilenceGenerator gen = new MP3SilenceGenerator(AudioSampleRate.SR_8000, MP3Setting.getDefault());
MP3SilenceGenerator gen = new MP3SilenceGenerator(InputAudioFormat.SR_8000, MP3Setting.getDefault());
List<byte[]> audio = gen.generate(173);

View File

@@ -31,16 +31,16 @@ public class MP3SilenceGenerator implements ISilenceGenerator
public static final boolean CONSTANT_BIT_RATE = false;
private MP3AudioConverter mMP3AudioConverter;
private AudioSampleRate mAudioSampleRate;
private InputAudioFormat mInputAudioFormat;
private byte[] mPreviousPartialFrameData;
/**
* Generates MP3 audio silence frames
*/
public MP3SilenceGenerator(AudioSampleRate audioSampleRate, MP3Setting setting)
public MP3SilenceGenerator(InputAudioFormat inputAudioFormat, MP3Setting setting)
{
mAudioSampleRate = audioSampleRate;
mMP3AudioConverter = new MP3AudioConverter(audioSampleRate, setting);
mInputAudioFormat = inputAudioFormat;
mMP3AudioConverter = new MP3AudioConverter(inputAudioFormat, setting);
}
/**
@@ -53,7 +53,7 @@ public class MP3SilenceGenerator implements ISilenceGenerator
double duration_secs = (double)duration_ms / 1000.0;
//We generate silence at 8000 kHz, because it gets resampled by the silence generator to the target rate
int length = (int)(duration_secs * AudioSampleRate.SR_8000.getSampleRate());
int length = (int)(duration_secs * InputAudioFormat.SR_8000.getSampleRate());
List<float[]> silenceBuffers = new ArrayList<>();
@@ -94,7 +94,7 @@ public class MP3SilenceGenerator implements ISilenceGenerator
{
mLog.debug("Starting ...");
MP3SilenceGenerator generator = new MP3SilenceGenerator(AudioSampleRate.SR_44100, MP3Setting.VBR_7);
MP3SilenceGenerator generator = new MP3SilenceGenerator(InputAudioFormat.SR_44100, MP3Setting.VBR_7);
int x = 100;