ulaw to wav pcm

Author: pablodz

audio/formats/pcm/pcm.go

@@ -2,62 +2,25 @@
 package pcm
 
 import (
-	"math"
+	"log"
 
 	"github.com/sopro-dev/sopro-core/audio"
 )
 
 type PCMFormat struct{}
 
 func (f *PCMFormat) Decode(data []byte, info audio.AudioInfo) []float64 {
-	sampleSize := info.BitDepth / 8
-	numSamples := len(data) / (sampleSize * info.Channels)
-	pcm := make([]float64, numSamples*info.Channels)
-
-	for i := 0; i < numSamples; i++ {
-		for ch := 0; ch < info.Channels; ch++ {
-			startIndex := (i*info.Channels + ch) * sampleSize
-			endIndex := startIndex + sampleSize
-			sample := int64(0)
-
-			for j := startIndex; j < endIndex; j++ {
-				sample |= int64(data[j]) << ((j - startIndex) * 8)
-			}
-
-			divider := math.Pow(2, float64(info.BitDepth-1))
-			if info.FloatFormat {
-				divider = math.Pow(2, float64(info.BitDepth))
-			}
-
-			pcm[i*info.Channels+ch] = float64(sample) / divider
-		}
-	}
-
-	return pcm
+	log.Printf("Not implemented")
+	return nil
 }
 
 func (f *PCMFormat) Encode(audioData []float64, info audio.AudioInfo) []byte {
-	sampleSize := info.BitDepth / 8
-	numSamples := len(audioData) / info.Channels
-	encoded := make([]byte, numSamples*info.Channels*sampleSize)
-
-	for i := 0; i < numSamples; i++ {
-		for ch := 0; ch < info.Channels; ch++ {
-			sampleIndex := i*info.Channels + ch
-			sample := audioData[sampleIndex]
-
-			sample = math.Max(-1.0, math.Min(sample, 1.0))
 
-			if !info.FloatFormat {
-				sample *= 0.5
-			}
-			sampleInt := int64((1 << (info.BitDepth - 1)) * int(sample))
-			for j := 0; j < sampleSize; j++ {
-				shift := uint(j * 8)
-				encoded[i*info.Channels*sampleSize+ch*sampleSize+j] = byte((sampleInt >> shift) & 0xFF)
-			}
-		}
+	// convert float64 to byte
+	data := make([]byte, len(audioData))
+	for i := 0; i < len(audioData); i++ {
+		data[i] = byte(audioData[i])
 	}
 
-	return encoded
+	return data
 }

audio/formats/ulaw/mulaw.go

@@ -0,0 +1,28 @@
+package mulaw
+
+import (
+	"log"
+
+	"github.com/sopro-dev/sopro-core/audio"
+	"github.com/sopro-dev/sopro-core/audio/utils"
+)
+
+type MuLawFormat struct{}
+
+func (f *MuLawFormat) Decode(data []byte, info audio.AudioInfo) []float64 {
+
+	pcmData := make([]float64, len(data))
+	for i := 0; i < len(data); i++ {
+		pcmData[i] = float64(utils.DecodeFromULaw(data[i]))
+	}
+
+	log.Println("[Bytes][Org]", data[0:100])
+	log.Println("[Bytes][Pcm]", pcmData[0:100])
+
+	return pcmData
+}
+
+func (f *MuLawFormat) Encode(audioData []float64, info audio.AudioInfo) []byte {
+	log.Printf("Not implemented")
+	return nil
+}

audio/formats/ulaw/ulaw.go

@@ -1,6 +1,9 @@
 package audio
 
-import "unsafe"
+import (
+	"errors"
+	"unsafe"
+)
 
 type Transcoder struct {
 	InputFormat  AudioFormat
@@ -14,9 +17,34 @@ func NewTranscoder(inputFormat, outputFormat AudioFormat) *Transcoder {
 	}
 }
 
-func (t *Transcoder) Transcode(inputData []byte, info AudioInfo) []byte {
+func (t *Transcoder) Transcode(inputData []byte, info AudioInfo) ([]byte, error) {
+	err := validateAudioInfo(info)
+	if err != nil {
+		return nil, err
+	}
 	audioData := t.InputFormat.Decode(inputData, info)
-	return t.OutputFormat.Encode(audioData, info)
+	return t.OutputFormat.Encode(audioData, info), nil
+}
+
+var (
+	errInvalidBitDepth    = errors.New("invalid bit depth")
+	errInvalidNumChannels = errors.New("invalid number of channels")
+	errInvalidSampleRate  = errors.New("invalid sample rate")
+)
+
+func validateAudioInfo(info AudioInfo) error {
+	if info.BitDepth != 8 && info.BitDepth != 16 && info.BitDepth != 24 && info.BitDepth != 32 {
+		return errInvalidBitDepth
+	}
+
+	if info.Channels < 1 || info.Channels > 2 {
+		return errInvalidNumChannels
+	}
+
+	if info.SampleRate < 1 {
+		return errInvalidSampleRate
+	}
+	return nil
 }
 
 // IMPORTANT: DO NOT REMOVE THIS LINE

audio/transcoder.go

@@ -0,0 +1,42 @@
+package audio
+
+import "testing"
+
+func TestValidateAudioInfo(t *testing.T) {
+	// Test case 1: Valid audio info
+	info := AudioInfo{BitDepth: 16, Channels: 2, SampleRate: 44100}
+	err := validateAudioInfo(info)
+	if err != nil {
+		t.Errorf("Unexpected error: %v", err)
+	}
+
+	// Test case 2: Invalid bit depth
+	info = AudioInfo{BitDepth: 5, Channels: 2, SampleRate: 44100}
+	err = validateAudioInfo(info)
+	if err == nil {
+		t.Error("Expected an error for invalid bit depth")
+	}
+	if err != errInvalidBitDepth {
+		t.Errorf("Expected error %v, got %v", errInvalidBitDepth, err)
+	}
+
+	// Test case 3: Invalid number of channels
+	info = AudioInfo{BitDepth: 16, Channels: 0, SampleRate: 44100}
+	err = validateAudioInfo(info)
+	if err == nil {
+		t.Error("Expected an error for invalid number of channels")
+	}
+	if err != errInvalidNumChannels {
+		t.Errorf("Expected error %v, got %v", errInvalidNumChannels, err)
+	}
+
+	// Test case 4: Invalid sample rate
+	info = AudioInfo{BitDepth: 16, Channels: 2, SampleRate: 0}
+	err = validateAudioInfo(info)
+	if err == nil {
+		t.Error("Expected an error for invalid sample rate")
+	}
+	if err != errInvalidSampleRate {
+		t.Errorf("Expected error %v, got %v", errInvalidSampleRate, err)
+	}
+}

audio/transcoder_test.go

@@ -0,0 +1,31 @@
+package utils
+
+import (
+	"encoding/binary"
+)
+
+// MergeSliceOfBytes merges multiple slices of bytes into a single slice
+func MergeSliceOfBytes(slices ...[]byte) []byte {
+	var result []byte
+	for _, s := range slices {
+		result = append(result, s...)
+	}
+	return result
+}
+
+// IntToBytes converts an unsigned integer to a little-endian byte slice
+func IntToBytes(i interface{}) []byte {
+	switch v := i.(type) {
+	case uint32:
+		buf := make([]byte, 4)
+		binary.LittleEndian.PutUint32(buf, v)
+		return buf
+	case uint16:
+		buf := make([]byte, 2)
+		binary.LittleEndian.PutUint16(buf, v)
+		return buf
+	default:
+
+	}
+	return nil
+}

audio/utils/bytes.go

@@ -0,0 +1,48 @@
+package utils
+
+import (
+	"bytes"
+	"testing"
+)
+
+func TestMergeSliceOfBytes(t *testing.T) {
+	slices := [][]byte{
+		[]byte("hello"),
+		[]byte(" "),
+		[]byte("world"),
+		[]byte(""),
+	}
+	expected := []byte("hello world")
+
+	result := MergeSliceOfBytes(slices...)
+
+	if !bytes.Equal(result, expected) {
+		t.Errorf("Expected %s but got %s", expected, result)
+	}
+}
+
+func TestIntToBytes(t *testing.T) {
+	tests := []struct {
+		input    interface{}
+		expected []byte
+	}{
+		{uint32(123456), []byte{64, 226, 1, 0}}, // 123456 in little-endian bytes
+		{uint16(0), []byte{0, 0}},               // 0 in little-endian bytes
+		{uint16(65535), []byte{255, 255}},       // 65535 in little-endian bytes
+		{uint16(1), []byte{1, 0}},               // 65536 in little-endian bytes
+		{uint32(2084), []byte{24, 8, 0, 0}},     // 65536 in little-endian bytes
+		// Add more test cases as needed
+	}
+
+	for _, test := range tests {
+		result := IntToBytes(test.input)
+		if !bytesEqual(result, test.expected) {
+			t.Errorf("For input %v, expected %v, but got %v", test.input, test.expected, result)
+		}
+	}
+}
+
+// bytesEqual compares two byte slices for equality
+func bytesEqual(a, b []byte) bool {
+	return len(a) == len(b) && (len(a) == 0 || (a[0] == b[0] && bytesEqual(a[1:], b[1:])))
+}

audio/utils/bytes_test.go

@@ -0,0 +1,37 @@
+package utils
+
+// ulawDecode is a lookup table for u-law to LPCM
+var ulawDecode = [256]int16{
+	-32124, -31100, -30076, -29052, -28028, -27004, -25980, -24956,
+	-23932, -22908, -21884, -20860, -19836, -18812, -17788, -16764,
+	-15996, -15484, -14972, -14460, -13948, -13436, -12924, -12412,
+	-11900, -11388, -10876, -10364, -9852, -9340, -8828, -8316,
+	-7932, -7676, -7420, -7164, -6908, -6652, -6396, -6140,
+	-5884, -5628, -5372, -5116, -4860, -4604, -4348, -4092,
+	-3900, -3772, -3644, -3516, -3388, -3260, -3132, -3004,
+	-2876, -2748, -2620, -2492, -2364, -2236, -2108, -1980,
+	-1884, -1820, -1756, -1692, -1628, -1564, -1500, -1436,
+	-1372, -1308, -1244, -1180, -1116, -1052, -988, -924,
+	-876, -844, -812, -780, -748, -716, -684, -652,
+	-620, -588, -556, -524, -492, -460, -428, -396,
+	-372, -356, -340, -324, -308, -292, -276, -260,
+	-244, -228, -212, -196, -180, -164, -148, -132,
+	-120, -112, -104, -96, -88, -80, -72, -64,
+	-56, -48, -40, -32, -24, -16, -8, 0,
+	32124, 31100, 30076, 29052, 28028, 27004, 25980, 24956,
+	23932, 22908, 21884, 20860, 19836, 18812, 17788, 16764,
+	15996, 15484, 14972, 14460, 13948, 13436, 12924, 12412,
+	11900, 11388, 10876, 10364, 9852, 9340, 8828, 8316,
+	7932, 7676, 7420, 7164, 6908, 6652, 6396, 6140,
+	5884, 5628, 5372, 5116, 4860, 4604, 4348, 4092,
+	3900, 3772, 3644, 3516, 3388, 3260, 3132, 3004,
+	2876, 2748, 2620, 2492, 2364, 2236, 2108, 1980,
+	1884, 1820, 1756, 1692, 1628, 1564, 1500, 1436,
+	1372, 1308, 1244, 1180, 1116, 1052, 988, 924,
+	876, 844, 812, 780, 748, 716, 684, 652,
+	620, 588, 556, 524, 492, 460, 428, 396,
+	372, 356, 340, 324, 308, 292, 276, 260,
+	244, 228, 212, 196, 180, 164, 148, 132,
+	120, 112, 104, 96, 88, 80, 72, 64,
+	56, 48, 40, 32, 24, 16, 8, 0,
+}

audio/utils/tables.go

@@ -1,34 +1,6 @@
 package utils
 
 func DecodeFromULaw(uLaw byte) int16 {
-	uLaw = ^uLaw
-	sign := int16(1)
-	if uLaw&0x80 != 0 {
-		sign = -1
-	}
-	exponent := int((uLaw >> 4) & 0x07)
-	mantissa := int(uLaw&0x0F) + 16
-	return sign * int16(mantissa) << uint(exponent+3)
-}
-
-func EncodeToULaw(sample int16) byte {
-	sign := byte(0)
-	if sample < 0 {
-		sign = 0x80
-		sample = -sample
-	}
-
-	sample = sample + 33
-	if sample > 0x7FFF {
-		sample = 0x7FFF
-	}
-
-	exponent := byte(7)
-	for (sample & 0x4000) == 0 {
-		exponent--
-		sample <<= 1
-	}
-
-	mantissa := byte(sample >> 6)
-	return ^byte(sign | (exponent << 4) | mantissa)
+	pcm := ulawDecode[uLaw]
+	return int16(pcm)
 }