Merged in audio-compression (pull request #142)

stream/adpcm & cmd/adpcm: Audio compression (ADPCM)

Approved-by: kortschak <dan@kortschak.io>
This commit is contained in:
Trek Hopton 2019-03-15 07:38:01 +00:00
commit adf58b800b
5 changed files with 568 additions and 0 deletions

View File

@ -5,6 +5,7 @@ av is a collection of tools and packages written in Go for audio-video processin
# Authors
Alan Noble
Saxon A. Nelson-Milton <saxon.milton@gmail.com>
Trek Hopton <trek@ausocean.org>
# Description

333
audio/adpcm/adpcm.go Normal file
View File

@ -0,0 +1,333 @@
/*
NAME
adpcm.go
DESCRIPTION
adpcm.go contains functions for encoding/compressing pcm into adpcm and decoding/decompressing back to pcm.
AUTHOR
Trek Hopton <trek@ausocean.org>
LICENSE
adpcm.go is Copyright (C) 2018 the Australian Ocean Lab (AusOcean)
It is free software: you can redistribute it and/or modify them
under the terms of the GNU General Public License as published by the
Free Software Foundation, either version 3 of the License, or (at your
option) any later version.
It is distributed in the hope that it will be useful, but WITHOUT
ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
for more details.
You should have received a copy of the GNU General Public License in gpl.txt.
If not, see [GNU licenses](http://www.gnu.org/licenses).
*/
/*
Original IMA/DVI ADPCM specification: (http://www.cs.columbia.edu/~hgs/audio/dvi/IMA_ADPCM.pdf).
Reference algorithms for ADPCM compression and decompression are in part 6.
*/
package adpcm
import (
"bytes"
"encoding/binary"
"fmt"
)
// encoder is used to encode to ADPCM from PCM data.
// pred and index hold state that persists between calls to encodeSample and calcHead.
// dest is the output buffer that implements io.writer and io.bytewriter, ie. where the encoded ADPCM data is written to.
type encoder struct {
dest *bytes.Buffer
pred int16
index int16
}
// decoder is used to decode from ADPCM to PCM data.
// pred, index, and step hold state that persists between calls to decodeSample.
// dest is the output buffer that implements io.writer and io.bytewriter, ie. where the decoded PCM data is written to.
type decoder struct {
dest *bytes.Buffer
pred int16
index int16
step int16
}
// PcmBS is the size of the blocks that an encoder uses.
// 'encodeBlock' will encode PcmBS bytes at a time and the output will be AdpcmBS bytes long.
const PcmBS = 1010
// AdpcmBS is the size of the blocks that a decoder uses.
// 'decodeBlock' will decode AdpcmBS bytes at a time and the output will be PcmBS bytes long.
const AdpcmBS = 256
// Table of index changes (see spec).
var indexTable = []int16{
-1, -1, -1, -1, 2, 4, 6, 8,
-1, -1, -1, -1, 2, 4, 6, 8,
}
// Quantize step size table (see spec).
var stepTable = []int16{
7, 8, 9, 10, 11, 12, 13, 14,
16, 17, 19, 21, 23, 25, 28, 31,
34, 37, 41, 45, 50, 55, 60, 66,
73, 80, 88, 97, 107, 118, 130, 143,
157, 173, 190, 209, 230, 253, 279, 307,
337, 371, 408, 449, 494, 544, 598, 658,
724, 796, 876, 963, 1060, 1166, 1282, 1411,
1552, 1707, 1878, 2066, 2272, 2499, 2749, 3024,
3327, 3660, 4026, 4428, 4871, 5358, 5894, 6484,
7132, 7845, 8630, 9493, 10442, 11487, 12635, 13899,
15289, 16818, 18500, 20350, 22385, 24623, 27086, 29794,
32767,
}
// NewEncoder retuns a new ADPCM encoder.
func NewEncoder(dst *bytes.Buffer) *encoder {
e := encoder{
dest: dst,
}
return &e
}
// NewDecoder retuns a new ADPCM decoder.
func NewDecoder(dst *bytes.Buffer) *decoder {
d := decoder{
step: stepTable[0],
dest: dst,
}
return &d
}
// encodeSample takes a single 16 bit PCM sample and
// returns a byte of which the last 4 bits are an encoded ADPCM nibble.
func (e *encoder) encodeSample(sample int16) byte {
// Find difference of actual sample from encoder's prediction.
delta := sample - e.pred
// Create and set sign bit for nibble and find absolute value of difference.
var nib byte
if delta < 0 {
nib = 8
delta = -delta
}
step := stepTable[e.index]
diff := step >> 3
var mask byte = 4
for i := 0; i < 3; i++ {
if delta > step {
nib |= mask
delta -= step
diff += step
}
mask >>= 1
step >>= 1
}
// Adjust predicted sample based on calculated difference.
if nib&8 != 0 {
e.pred -= diff
} else {
e.pred += diff
}
e.index += indexTable[nib&7]
// Check for underflow and overflow.
if e.index < 0 {
e.index = 0
} else if e.index > int16(len(stepTable)-1) {
e.index = int16(len(stepTable) - 1)
}
return nib
}
// decodeSample takes a byte, the last 4 bits of which contain a single
// 4 bit ADPCM nibble, and returns a 16 bit decoded PCM sample.
func (d *decoder) decodeSample(nibble byte) int16 {
// Calculate difference.
var diff int16
if nibble&4 != 0 {
diff += d.step
}
if nibble&2 != 0 {
diff += d.step >> 1
}
if nibble&1 != 0 {
diff += d.step >> 2
}
diff += d.step >> 3
// Account for sign bit.
if nibble&8 != 0 {
diff = -diff
}
// Adjust predicted sample based on calculated difference.
d.pred += diff
// Adjust index into step size lookup table using nibble.
d.index += indexTable[nibble]
// Check for overflow and underflow.
if d.index < 0 {
d.index = 0
} else if d.index > int16(len(stepTable)-1) {
d.index = int16(len(stepTable) - 1)
}
// Find new quantizer step size.
d.step = stepTable[d.index]
return d.pred
}
// calcHead sets the state for the encoder by running the first sample through
// the encoder, and writing the first sample to the encoder's io.Writer (dest).
// It returns the number of bytes written to the encoder's io.Writer (dest) along with any errors.
func (e *encoder) calcHead(sample []byte) (int, error) {
// Check that we are given 1 16-bit sample (2 bytes).
const sampSize = 2
if len(sample) != sampSize {
return 0, fmt.Errorf("length of given byte array is: %v, expected: %v", len(sample), sampSize)
}
intSample := int16(binary.LittleEndian.Uint16(sample))
e.encodeSample(intSample)
n, err := e.dest.Write(sample)
if err != nil {
return n, err
}
err = e.dest.WriteByte(byte(uint16(e.index)))
if err != nil {
return n, err
}
n++
err = e.dest.WriteByte(byte(0x00))
if err != nil {
return n, err
}
n++
return n, nil
}
// encodeBlock takes a slice of 1010 bytes (505 16-bit PCM samples).
// It writes encoded (compressed) bytes (each byte containing two ADPCM nibbles) to the encoder's io.Writer (dest).
// The number of bytes written is returned along with any errors.
// Note: nibbles are output in little endian order, eg. n1n0 n3n2 n5n4...
// Note: first 4 bytes are for initializing the decoder before decoding a block.
// - First two bytes contain the first 16-bit sample uncompressed.
// - Third byte is the decoder's starting index for the block, the fourth is padding and ignored.
func (e *encoder) encodeBlock(block []byte) (int, error) {
if len(block) != PcmBS {
return 0, fmt.Errorf("unsupported block size. Given: %v, expected: %v, ie. 505 16-bit PCM samples", len(block), PcmBS)
}
n, err := e.calcHead(block[0:2])
if err != nil {
return n, err
}
for i := 3; i < PcmBS; i += 4 {
nib1 := e.encodeSample(int16(binary.LittleEndian.Uint16(block[i-1 : i+1])))
nib2 := e.encodeSample(int16(binary.LittleEndian.Uint16(block[i+1 : i+3])))
err = e.dest.WriteByte(byte((nib2 << 4) | nib1))
if err != nil {
return n, err
}
n++
}
return n, nil
}
// decodeBlock takes a slice of 256 bytes, each byte after the first 4 should contain two ADPCM encoded nibbles.
// It writes the resulting decoded (decompressed) 16-bit PCM samples to the decoder's io.Writer (dest).
// The number of bytes written is returned along with any errors.
func (d *decoder) decodeBlock(block []byte) (int, error) {
if len(block) != AdpcmBS {
return 0, fmt.Errorf("unsupported block size. Given: %v, expected: %v", len(block), AdpcmBS)
}
// Initialize decoder with first 4 bytes of the block.
d.pred = int16(binary.LittleEndian.Uint16(block[0:2]))
d.index = int16(block[2])
d.step = stepTable[d.index]
n, err := d.dest.Write(block[0:2])
if err != nil {
return n, err
}
// For each byte, seperate it into two nibbles (each nibble is a compressed sample),
// then decode each nibble and output the resulting 16-bit samples.
for i := 4; i < AdpcmBS; i++ {
twoNibs := block[i]
nib2 := byte(twoNibs >> 4)
nib1 := byte((nib2 << 4) ^ twoNibs)
firstBytes := make([]byte, 2)
binary.LittleEndian.PutUint16(firstBytes, uint16(d.decodeSample(nib1)))
_n, err := d.dest.Write(firstBytes)
n += _n
if err != nil {
return n, err
}
secondBytes := make([]byte, 2)
binary.LittleEndian.PutUint16(secondBytes, uint16(d.decodeSample(nib2)))
_n, err = d.dest.Write(secondBytes)
n += _n
if err != nil {
return n, err
}
}
return n, nil
}
// Write takes a slice of bytes of arbitrary length representing pcm and encodes in into adpcm.
// It writes its output to the encoder's dest.
// The number of bytes written out is returned along with any error that occured.
func (e *encoder) Write(inPcm []byte) (int, error) {
numBlocks := len(inPcm) / PcmBS
n := 0
for i := 0; i < numBlocks; i++ {
block := inPcm[PcmBS*i : PcmBS*(i+1)]
_n, err := e.encodeBlock(block)
n += _n
if err != nil {
return n, err
}
}
return n, nil
}
// Write takes a slice of bytes of arbitrary length representing adpcm and decodes in into pcm.
// It writes its output to the decoder's dest.
// The number of bytes written out is returned along with any error that occured.
func (d *decoder) Write(inAdpcm []byte) (int, error) {
numBlocks := len(inAdpcm) / AdpcmBS
n := 0
for i := 0; i < numBlocks; i++ {
block := inAdpcm[AdpcmBS*i : AdpcmBS*(i+1)]
_n, err := d.decodeBlock(block)
n += _n
if err != nil {
return n, err
}
}
return n, nil
}

92
audio/adpcm/adpcm_test.go Normal file
View File

@ -0,0 +1,92 @@
/*
NAME
adpcm_test.go
DESCRIPTION
adpcm_test.go contains tests for the adpcm package.
AUTHOR
Trek Hopton <trek@ausocean.org>
LICENSE
adpcm_test.go is Copyright (C) 2018 the Australian Ocean Lab (AusOcean)
It is free software: you can redistribute it and/or modify them
under the terms of the GNU General Public License as published by the
Free Software Foundation, either version 3 of the License, or (at your
option) any later version.
It is distributed in the hope that it will be useful, but WITHOUT
ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
for more details.
You should have received a copy of the GNU General Public License in gpl.txt.
If not, see [GNU licenses](http://www.gnu.org/licenses).
*/
package adpcm
import (
"bytes"
"io/ioutil"
"testing"
)
// TestEncodeBlock will read PCM data, encode it in blocks and generate ADPCM
// then compare the result with expected ADPCM.
func TestEncodeBlock(t *testing.T) {
// Read input pcm.
pcm, err := ioutil.ReadFile("../../../test/test-data/av/input/raw-voice.pcm")
if err != nil {
t.Errorf("Unable to read input PCM file: %v", err)
}
// Encode adpcm.
numBlocks := len(pcm) / PcmBS
comp := bytes.NewBuffer(make([]byte, 0, AdpcmBS*numBlocks))
enc := NewEncoder(comp)
_, err = enc.Write(pcm)
if err != nil {
t.Errorf("Unable to write to encoder: %v", err)
}
// Read expected adpcm file.
exp, err := ioutil.ReadFile("../../../test/test-data/av/output/encoded-voice.adpcm")
if err != nil {
t.Errorf("Unable to read expected ADPCM file: %v", err)
}
if !bytes.Equal(comp.Bytes(), exp) {
t.Error("ADPCM generated does not match expected ADPCM")
}
}
// TestDecodeBlock will read encoded ADPCM, decode it in blocks and then compare the
// resulting PCM with the expected decoded PCM.
func TestDecodeBlock(t *testing.T) {
// Read adpcm.
comp, err := ioutil.ReadFile("../../../test/test-data/av/input/encoded-voice.adpcm")
if err != nil {
t.Errorf("Unable to read input ADPCM file: %v", err)
}
// Decode adpcm.
numBlocks := len(comp) / AdpcmBS
decoded := bytes.NewBuffer(make([]byte, 0, PcmBS*numBlocks))
dec := NewDecoder(decoded)
_, err = dec.Write(comp)
if err != nil {
t.Errorf("Unable to write to decoder: %v", err)
}
// Read expected pcm file.
exp, err := ioutil.ReadFile("../../../test/test-data/av/output/decoded-voice.pcm")
if err != nil {
t.Errorf("Unable to read expected PCM file: %v", err)
}
if !bytes.Equal(decoded.Bytes(), exp) {
t.Error("PCM generated does not match expected PCM")
}
}

View File

@ -0,0 +1,71 @@
/*
NAME
decode-pcm.go
DESCRIPTION
decode-pcm.go is a program for decoding/decompressing an adpcm file to a pcm file.
AUTHOR
Trek Hopton <trek@ausocean.org>
LICENSE
decode-pcm.go is Copyright (C) 2018 the Australian Ocean Lab (AusOcean)
It is free software: you can redistribute it and/or modify them
under the terms of the GNU General Public License as published by the
Free Software Foundation, either version 3 of the License, or (at your
option) any later version.
It is distributed in the hope that it will be useful, but WITHOUT
ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
for more details.
You should have received a copy of the GNU General Public License in gpl.txt.
If not, see [GNU licenses](http://www.gnu.org/licenses).
*/
package main
import (
"bytes"
"flag"
"fmt"
"io/ioutil"
"log"
"bitbucket.org/ausocean/av/audio/adpcm"
)
// This program accepts an input file encoded in adpcm and outputs a decoded pcm file.
// Input and output file names can be specified as arguments.
func main() {
var inPath string
var outPath string
flag.StringVar(&inPath, "in", "encoded.adpcm", "file path of input")
flag.StringVar(&outPath, "out", "decoded.pcm", "file path of output data")
flag.Parse()
// Read adpcm.
comp, err := ioutil.ReadFile(inPath)
if err != nil {
log.Fatal(err)
}
fmt.Println("Read", len(comp), "bytes from file", inPath)
// Decode adpcm.
numBlocks := len(comp) / adpcm.AdpcmBS
decoded := bytes.NewBuffer(make([]byte, 0, adpcm.PcmBS*numBlocks))
dec := adpcm.NewDecoder(decoded)
_, err = dec.Write(comp)
if err != nil {
log.Fatal(err)
}
// Save pcm to file.
err = ioutil.WriteFile(outPath, decoded.Bytes(), 0644)
if err != nil {
log.Fatal(err)
}
fmt.Println("Decoded and wrote", len(decoded.Bytes()), "bytes to file", outPath)
}

View File

@ -0,0 +1,71 @@
/*
NAME
encode-pcm.go
DESCRIPTION
encode-pcm.go is a program for encoding/compressing a pcm file to an adpcm file.
AUTHOR
Trek Hopton <trek@ausocean.org>
LICENSE
encode-pcm.go is Copyright (C) 2018 the Australian Ocean Lab (AusOcean)
It is free software: you can redistribute it and/or modify them
under the terms of the GNU General Public License as published by the
Free Software Foundation, either version 3 of the License, or (at your
option) any later version.
It is distributed in the hope that it will be useful, but WITHOUT
ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
for more details.
You should have received a copy of the GNU General Public License in gpl.txt.
If not, see [GNU licenses](http://www.gnu.org/licenses).
*/
package main
import (
"bytes"
"flag"
"fmt"
"io/ioutil"
"log"
"bitbucket.org/ausocean/av/audio/adpcm"
)
// This program accepts an input pcm file and outputs an encoded adpcm file.
// Input and output file names can be specified as arguments.
func main() {
var inPath string
var adpcmPath string
flag.StringVar(&inPath, "in", "data.pcm", "file path of input data")
flag.StringVar(&adpcmPath, "out", "encoded.adpcm", "file path of output")
flag.Parse()
// Read pcm.
pcm, err := ioutil.ReadFile(inPath)
if err != nil {
log.Fatal(err)
}
fmt.Println("Read", len(pcm), "bytes from file", inPath)
// Encode adpcm.
numBlocks := len(pcm) / adpcm.PcmBS
comp := bytes.NewBuffer(make([]byte, 0, adpcm.AdpcmBS*numBlocks))
enc := adpcm.NewEncoder(comp)
_, err = enc.Write(pcm)
if err != nil {
log.Fatal(err)
}
// Save adpcm to file.
err = ioutil.WriteFile(adpcmPath, comp.Bytes(), 0644)
if err != nil {
log.Fatal(err)
}
fmt.Println("Encoded and wrote", len(comp.Bytes()), "bytes to file", adpcmPath)
}