add RTP/MPEG-2 audio decoder and encoder (#257)

README.md

@@ -45,7 +45,7 @@ Features:
   * Parse RTSP elements
   * Encode/decode format-specific frames into/from RTP packets. The following formats are supported:
     * Video: H264, H265, M-JPEG, VP8, VP9
-    * Audio: G711 (PCMA, PCMU), G722, LPCM, MPEG4 Audio (AAC), Opus
+    * Audio: G711 (PCMA, PCMU), G722, LPCM, MPEG-2 Audio (MP3), MPEG-4 Audio (AAC), Opus
 
 ## Table of contents
 

go.mod

@@ -4,7 +4,7 @@ go 1.18
 
 require (
 	github.com/asticode/go-astits v1.11.0
-	github.com/bluenviron/mediacommon v0.3.1
+	github.com/bluenviron/mediacommon v0.4.1
 	github.com/google/uuid v1.3.0
 	github.com/pion/rtcp v1.2.10
 	github.com/pion/rtp v0.0.0-20230107162714-c3ea6851e25b

go.sum

@@ -2,8 +2,8 @@ github.com/asticode/go-astikit v0.30.0 h1:DkBkRQRIxYcknlaU7W7ksNfn4gMFsB0tqMJflx
 github.com/asticode/go-astikit v0.30.0/go.mod h1:h4ly7idim1tNhaVkdVBeXQZEE3L0xblP7fCWbgwipF0=
 github.com/asticode/go-astits v1.11.0 h1:GTHUXht0ZXAJXsVbsLIcyfHr1Bchi4QQwMARw2ZWAng=
 github.com/asticode/go-astits v1.11.0/go.mod h1:QSHmknZ51pf6KJdHKZHJTLlMegIrhega3LPWz3ND/iI=
-github.com/bluenviron/mediacommon v0.3.1 h1:C4okNqyN1Mg5CVGcGKk2tEk9Uj2hHZusHV7nqdjn1Lk=
+github.com/bluenviron/mediacommon v0.4.1 h1:oiqvqwnZ0NbB+mCZjuyBtgY7cF88rZdhb/PTSNG9M+Q=
-github.com/bluenviron/mediacommon v0.3.1/go.mod h1:t0dqPsWUTchyvib0MhixIwXEgvDX4V9G+I0GzWLQRb8=
+github.com/bluenviron/mediacommon v0.4.1/go.mod h1:t0dqPsWUTchyvib0MhixIwXEgvDX4V9G+I0GzWLQRb8=
 github.com/davecgh/go-spew v1.1.0/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
 github.com/davecgh/go-spew v1.1.1 h1:vj9j/u1bqnvCEfJOwUhtlOARqs3+rkHYY13jYWTU97c=
 github.com/davecgh/go-spew v1.1.1/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=

pkg/formats/mjpeg.go

@@ -1,4 +1,4 @@
-package formats
+package formats //nolint:dupl
 
 import (
 	"github.com/pion/rtp"

pkg/formats/mpeg2_audio.go

@@ -1,7 +1,9 @@
-package formats
+package formats //nolint:dupl
 
 import (
 	"github.com/pion/rtp"
+
+	"github.com/bluenviron/gortsplib/v3/pkg/formats/rtpmpeg2audio"
 )
 
 // MPEG2Audio is a RTP format that uses a MPEG-1 or MPEG-2 audio codec.
@@ -44,3 +46,17 @@ func (f *MPEG2Audio) FMTP() map[string]string {
 func (f *MPEG2Audio) PTSEqualsDTS(*rtp.Packet) bool {
 	return true
 }
+
+// CreateDecoder creates a decoder able to decode the content of the format.
+func (f *MPEG2Audio) CreateDecoder() *rtpmpeg2audio.Decoder {
+	d := &rtpmpeg2audio.Decoder{}
+	d.Init()
+	return d
+}
+
+// CreateEncoder creates an encoder able to encode the content of the format.
+func (f *MPEG2Audio) CreateEncoder() *rtpmpeg2audio.Encoder {
+	e := &rtpmpeg2audio.Encoder{}
+	e.Init()
+	return e
+}

pkg/formats/mpeg2_audio_test.go

@@ -13,3 +13,43 @@ func TestMPEG2AudioAttributes(t *testing.T) {
 	require.Equal(t, 90000, format.ClockRate())
 	require.Equal(t, true, format.PTSEqualsDTS(&rtp.Packet{}))
 }
+
+func TestMPEG2AudioDecEncoder(t *testing.T) {
+	format := &MPEG2Audio{}
+
+	enc := format.CreateEncoder()
+	pkts, err := enc.Encode([][]byte{{
+		0xff, 0xfb, 0x14, 0x64, 0x00, 0x0f, 0xf0, 0x00,
+		0x00, 0x69, 0x00, 0x00, 0x00, 0x08, 0x00, 0x00,
+		0x0d, 0x20, 0x00, 0x00, 0x01, 0x00, 0x00, 0x01,
+		0xa4, 0x00, 0x00, 0x00, 0x20, 0x00, 0x00, 0x34,
+		0x80, 0x00, 0x00, 0x04, 0x4c, 0x41, 0x4d, 0x45,
+		0x33, 0x2e, 0x31, 0x30, 0x30, 0x55, 0x55, 0x55,
+		0x55, 0x55, 0x55, 0x55, 0x55, 0x55, 0x55, 0x55,
+		0x55, 0x55, 0x55, 0x55, 0x55, 0x55, 0x55, 0x55,
+		0x55, 0xc0, 0x65, 0xf4, 0xa0, 0x31, 0x8f, 0xce,
+		0x8d, 0x46, 0xfc, 0x8c, 0x73, 0xb9, 0x34, 0x3e,
+		0xb5, 0x03, 0x39, 0xc0, 0x04, 0x01, 0x98, 0x44,
+		0x38, 0xe0, 0x98, 0x10, 0x9b, 0xa8, 0x0f, 0xa8,
+	}}, 0)
+	require.NoError(t, err)
+	require.Equal(t, format.PayloadType(), pkts[0].PayloadType)
+
+	dec := format.CreateDecoder()
+	byts, _, err := dec.Decode(pkts[0])
+	require.NoError(t, err)
+	require.Equal(t, [][]byte{{
+		0xff, 0xfb, 0x14, 0x64, 0x00, 0x0f, 0xf0, 0x00,
+		0x00, 0x69, 0x00, 0x00, 0x00, 0x08, 0x00, 0x00,
+		0x0d, 0x20, 0x00, 0x00, 0x01, 0x00, 0x00, 0x01,
+		0xa4, 0x00, 0x00, 0x00, 0x20, 0x00, 0x00, 0x34,
+		0x80, 0x00, 0x00, 0x04, 0x4c, 0x41, 0x4d, 0x45,
+		0x33, 0x2e, 0x31, 0x30, 0x30, 0x55, 0x55, 0x55,
+		0x55, 0x55, 0x55, 0x55, 0x55, 0x55, 0x55, 0x55,
+		0x55, 0x55, 0x55, 0x55, 0x55, 0x55, 0x55, 0x55,
+		0x55, 0xc0, 0x65, 0xf4, 0xa0, 0x31, 0x8f, 0xce,
+		0x8d, 0x46, 0xfc, 0x8c, 0x73, 0xb9, 0x34, 0x3e,
+		0xb5, 0x03, 0x39, 0xc0, 0x04, 0x01, 0x98, 0x44,
+		0x38, 0xe0, 0x98, 0x10, 0x9b, 0xa8, 0x0f, 0xa8,
+	}}, byts)
+}

pkg/formats/rtph264/decoder.go

@@ -23,6 +23,7 @@ var ErrNonStartingPacketAndNoPrevious = errors.New(
 	"received a non-starting fragment without any previous starting fragment")
 
 // Decoder is a RTP/H264 decoder.
+// Specification: https://datatracker.ietf.org/doc/html/rfc6184
 type Decoder struct {
 	// indicates the packetization mode.
 	PacketizationMode int

pkg/formats/rtph264/decoder_test.go

@@ -204,7 +204,7 @@ func TestDecodeUntilMarker(t *testing.T) {
 	require.Equal(t, [][]byte{{0x01, 0x02}, {0x01, 0x02}}, nalus)
 }
 
-func FuzzDecoderUnmarshal(f *testing.F) {
+func FuzzDecoder(f *testing.F) {
 	d := &Decoder{}
 	d.Init()
 

pkg/formats/rtph264/encoder.go

@@ -22,6 +22,7 @@ func randUint32() uint32 {
 }
 
 // Encoder is a RTP/H264 encoder.
+// Specification: https://datatracker.ietf.org/doc/html/rfc6184
 type Encoder struct {
 	// payload type of packets.
 	PayloadType uint8

pkg/formats/rtph265/decoder.go

@@ -22,6 +22,7 @@ var ErrNonStartingPacketAndNoPrevious = errors.New(
 	"received a non-starting fragment without any previous starting fragment")
 
 // Decoder is a RTP/H265 decoder.
+// Specification: https://datatracker.ietf.org/doc/html/rfc7798
 type Decoder struct {
 	// indicates that NALUs have an additional field that specifies the decoding order.
 	MaxDONDiff int

pkg/formats/rtph265/decoder_test.go

@@ -56,7 +56,7 @@ func TestDecode(t *testing.T) {
 	}
 }
 
-func FuzzDecoderUnmarshal(f *testing.F) {
+func FuzzDecoder(f *testing.F) {
 	d := &Decoder{}
 	d.Init()
 

pkg/formats/rtph265/encoder.go

@@ -21,6 +21,7 @@ func randUint32() uint32 {
 }
 
 // Encoder is a RTP/H265 encoder.
+// Specification: https://datatracker.ietf.org/doc/html/rfc7798
 type Encoder struct {
 	// payload type of packets.
 	PayloadType uint8

pkg/formats/rtplpcm/decoder.go

@@ -10,6 +10,7 @@ import (
 )
 
 // Decoder is a RTP/LPCM decoder.
+// Specification: https://datatracker.ietf.org/doc/html/rfc3190
 type Decoder struct {
 	BitDepth     int
 	SampleRate   int

pkg/formats/rtplpcm/decoder_test.go

@@ -54,7 +54,7 @@ func TestDecode(t *testing.T) {
 	}
 }
 
-func FuzzDecoderUnmarshal(f *testing.F) {
+func FuzzDecoder(f *testing.F) {
 	d := &Decoder{
 		BitDepth:     24,
 		SampleRate:   48000,

pkg/formats/rtplpcm/encoder.go

@@ -21,6 +21,7 @@ func randUint32() uint32 {
 }
 
 // Encoder is a RTP/LPCM encoder.
+// Specification: https://datatracker.ietf.org/doc/html/rfc3190
 type Encoder struct {
 	// payload type of packets.
 	PayloadType uint8

pkg/formats/rtpmjpeg/decoder.go

@@ -95,6 +95,7 @@ var chmAcSymbols = []byte{ //nolint:dupl
 }
 
 // Decoder is a RTP/M-JPEG decoder.
+// Specification: https://datatracker.ietf.org/doc/html/rfc2435
 type Decoder struct {
 	timeDecoder         *rtptime.Decoder
 	firstPacketReceived bool

pkg/formats/rtpmjpeg/decoder_test.go

@@ -30,7 +30,7 @@ func TestDecode(t *testing.T) {
 	}
 }
 
-func FuzzDecoderUnmarshal(f *testing.F) {
+func FuzzDecoder(f *testing.F) {
 	d := &Decoder{}
 	d.Init()
 

pkg/formats/rtpmjpeg/encoder.go

@@ -24,6 +24,7 @@ func randUint32() uint32 {
 }
 
 // Encoder is a RTP/M-JPEG encoder.
+// Specification: https://datatracker.ietf.org/doc/html/rfc2435
 type Encoder struct {
 	// SSRC of packets (optional).
 	// It defaults to a random value.

pkg/formats/rtpmpeg2audio/decoder.go

@@ -0,0 +1,46 @@
+package rtpmpeg2audio
+
+import (
+	"fmt"
+	"time"
+
+	"github.com/bluenviron/mediacommon/pkg/codecs/mpeg2audio"
+	"github.com/pion/rtp"
+
+	"github.com/bluenviron/gortsplib/v3/pkg/rtptime"
+)
+
+// Decoder is a RTP/MPEG2-audio decoder.
+// Specification: https://datatracker.ietf.org/doc/html/rfc2250
+type Decoder struct {
+	timeDecoder *rtptime.Decoder
+}
+
+// Init initializes the decoder.
+func (d *Decoder) Init() {
+	d.timeDecoder = rtptime.NewDecoder(90000)
+}
+
+// Decode decodes frames from a RTP/MPEG2-audio packet.
+func (d *Decoder) Decode(pkt *rtp.Packet) ([][]byte, time.Duration, error) {
+	if len(pkt.Payload) < 5 {
+		return nil, 0, fmt.Errorf("payload is too short")
+	}
+
+	mbz := uint16(pkt.Payload[0])<<8 | uint16(pkt.Payload[1])
+	if mbz != 0 {
+		return nil, 0, fmt.Errorf("invalid MBZ: %v", mbz)
+	}
+
+	offset := uint16(pkt.Payload[2])<<8 | uint16(pkt.Payload[3])
+	if offset != 0 {
+		return nil, 0, fmt.Errorf("fragmented units are not supported")
+	}
+
+	frames, err := mpeg2audio.SplitFrames(pkt.Payload[4:])
+	if err != nil {
+		return nil, 0, err
+	}
+
+	return frames, d.timeDecoder.Decode(pkt.Timestamp), nil
+}

pkg/formats/rtpmpeg2audio/decoder_test.go

@@ -0,0 +1,39 @@
+package rtpmpeg2audio
+
+import (
+	"testing"
+
+	"github.com/pion/rtp"
+	"github.com/stretchr/testify/require"
+)
+
+func TestDecode(t *testing.T) {
+	for _, ca := range cases {
+		t.Run(ca.name, func(t *testing.T) {
+			d := &Decoder{}
+			d.Init()
+
+			frames, _, err := d.Decode(ca.pkt)
+			require.NoError(t, err)
+			require.Equal(t, ca.frames, frames)
+		})
+	}
+}
+
+func FuzzDecoder(f *testing.F) {
+	d := &Decoder{}
+	d.Init()
+
+	f.Fuzz(func(t *testing.T, b []byte) {
+		d.Decode(&rtp.Packet{
+			Header: rtp.Header{
+				Version:        2,
+				PayloadType:    96,
+				SequenceNumber: 17645,
+				Timestamp:      2289527317,
+				SSRC:           0x9dbb7812,
+			},
+			Payload: b,
+		})
+	})
+}

pkg/formats/rtpmpeg2audio/encoder.go

@@ -0,0 +1,151 @@
+package rtpmpeg2audio
+
+import (
+	"crypto/rand"
+	"fmt"
+	"time"
+
+	"github.com/bluenviron/mediacommon/pkg/codecs/mpeg2audio"
+	"github.com/pion/rtp"
+
+	"github.com/bluenviron/gortsplib/v3/pkg/rtptime"
+)
+
+const (
+	rtpVersion = 2
+)
+
+func randUint32() uint32 {
+	var b [4]byte
+	rand.Read(b[:])
+	return uint32(b[0])<<24 | uint32(b[1])<<16 | uint32(b[2])<<8 | uint32(b[3])
+}
+
+func lenAggregated(frames [][]byte, frame []byte) int {
+	l := 4 + len(frame)
+	for _, fr := range frames {
+		l += len(fr)
+	}
+	return l
+}
+
+// Encoder is a RTP/MPEG2-audio encoder.
+// Specification: https://datatracker.ietf.org/doc/html/rfc2250
+type Encoder struct {
+	// SSRC of packets (optional).
+	// It defaults to a random value.
+	SSRC *uint32
+
+	// initial sequence number of packets (optional).
+	// It defaults to a random value.
+	InitialSequenceNumber *uint16
+
+	// initial timestamp of packets (optional).
+	// It defaults to a random value.
+	InitialTimestamp *uint32
+
+	// maximum size of packet payloads (optional).
+	// It defaults to 1460.
+	PayloadMaxSize int
+
+	sequenceNumber uint16
+	timeEncoder    *rtptime.Encoder
+}
+
+// Init initializes the encoder.
+func (e *Encoder) Init() {
+	if e.SSRC == nil {
+		v := randUint32()
+		e.SSRC = &v
+	}
+	if e.InitialSequenceNumber == nil {
+		v := uint16(randUint32())
+		e.InitialSequenceNumber = &v
+	}
+	if e.InitialTimestamp == nil {
+		v := randUint32()
+		e.InitialTimestamp = &v
+	}
+	if e.PayloadMaxSize == 0 {
+		e.PayloadMaxSize = 1460 // 1500 (UDP MTU) - 20 (IP header) - 8 (UDP header) - 12 (RTP header)
+	}
+
+	e.sequenceNumber = *e.InitialSequenceNumber
+	e.timeEncoder = rtptime.NewEncoder(90000, *e.InitialTimestamp)
+}
+
+// Encode encodes frames into RTP/MPEG2-audio packets.
+func (e *Encoder) Encode(frames [][]byte, pts time.Duration) ([]*rtp.Packet, error) {
+	var rets []*rtp.Packet
+	var batch [][]byte
+
+	for _, frame := range frames {
+		if len(frame) > e.PayloadMaxSize {
+			return nil, fmt.Errorf("frame is too big")
+		}
+
+		if lenAggregated(batch, frame) <= e.PayloadMaxSize {
+			batch = append(batch, frame)
+		} else {
+			// write last batch
+			if batch != nil {
+				pkt, err := e.writeBatch(batch, pts)
+				if err != nil {
+					return nil, err
+				}
+				rets = append(rets, pkt)
+
+				for _, frame := range batch {
+					var h mpeg2audio.FrameHeader
+					err := h.Unmarshal(frame)
+					if err != nil {
+						return nil, err
+					}
+
+					pts += time.Duration(h.SampleCount()) * time.Second / time.Duration(h.SampleRate)
+				}
+			}
+
+			// initialize new batch
+			batch = [][]byte{frame}
+		}
+	}
+
+	// write last batch
+	pkt, err := e.writeBatch(batch, pts)
+	if err != nil {
+		return nil, err
+	}
+	rets = append(rets, pkt)
+
+	return rets, nil
+}
+
+func (e *Encoder) writeBatch(frames [][]byte, pts time.Duration) (*rtp.Packet, error) {
+	l := 4
+	for _, frame := range frames {
+		l += len(frame)
+	}
+
+	payload := make([]byte, l)
+	n := 4
+	for _, frame := range frames {
+		n += copy(payload[n:], frame)
+	}
+
+	pkt := &rtp.Packet{
+		Header: rtp.Header{
+			Version:        rtpVersion,
+			PayloadType:    14,
+			SequenceNumber: e.sequenceNumber,
+			Timestamp:      e.timeEncoder.Encode(pts),
+			SSRC:           *e.SSRC,
+			Marker:         true,
+		},
+		Payload: payload,
+	}
+
+	e.sequenceNumber++
+
+	return pkt, nil
+}

pkg/formats/rtpmpeg2audio/encoder_test.go

@@ -0,0 +1,163 @@
+package rtpmpeg2audio
+
+import (
+	"testing"
+
+	"github.com/pion/rtp"
+	"github.com/stretchr/testify/require"
+)
+
+var cases = []struct {
+	name   string
+	frames [][]byte
+	pkt    *rtp.Packet
+}{
+	{
+		"single",
+		[][]byte{
+			{
+				0xff, 0xfb, 0x14, 0x64, 0x00, 0x0f, 0xf0, 0x00,
+				0x00, 0x69, 0x00, 0x00, 0x00, 0x08, 0x00, 0x00,
+				0x0d, 0x20, 0x00, 0x00, 0x01, 0x00, 0x00, 0x01,
+				0xa4, 0x00, 0x00, 0x00, 0x20, 0x00, 0x00, 0x34,
+				0x80, 0x00, 0x00, 0x04, 0x4c, 0x41, 0x4d, 0x45,
+				0x33, 0x2e, 0x31, 0x30, 0x30, 0x55, 0x55, 0x55,
+				0x55, 0x55, 0x55, 0x55, 0x55, 0x55, 0x55, 0x55,
+				0x55, 0x55, 0x55, 0x55, 0x55, 0x55, 0x55, 0x55,
+				0x55, 0xc0, 0x65, 0xf4, 0xa0, 0x31, 0x8f, 0xce,
+				0x8d, 0x46, 0xfc, 0x8c, 0x73, 0xb9, 0x34, 0x3e,
+				0xb5, 0x03, 0x39, 0xc0, 0x04, 0x01, 0x98, 0x44,
+				0x38, 0xe0, 0x98, 0x10, 0x9b, 0xa8, 0x0f, 0xa8,
+			},
+		},
+		&rtp.Packet{
+			Header: rtp.Header{
+				Version:        2,
+				Marker:         true,
+				PayloadType:    14,
+				SequenceNumber: 17645,
+				Timestamp:      2289526357,
+				SSRC:           0x9dbb7812,
+			},
+			Payload: []byte{
+				0x00, 0x00, 0x00, 0x00,
+				0xff, 0xfb, 0x14, 0x64, 0x00, 0x0f, 0xf0, 0x00,
+				0x00, 0x69, 0x00, 0x00, 0x00, 0x08, 0x00, 0x00,
+				0x0d, 0x20, 0x00, 0x00, 0x01, 0x00, 0x00, 0x01,
+				0xa4, 0x00, 0x00, 0x00, 0x20, 0x00, 0x00, 0x34,
+				0x80, 0x00, 0x00, 0x04, 0x4c, 0x41, 0x4d, 0x45,
+				0x33, 0x2e, 0x31, 0x30, 0x30, 0x55, 0x55, 0x55,
+				0x55, 0x55, 0x55, 0x55, 0x55, 0x55, 0x55, 0x55,
+				0x55, 0x55, 0x55, 0x55, 0x55, 0x55, 0x55, 0x55,
+				0x55, 0xc0, 0x65, 0xf4, 0xa0, 0x31, 0x8f, 0xce,
+				0x8d, 0x46, 0xfc, 0x8c, 0x73, 0xb9, 0x34, 0x3e,
+				0xb5, 0x03, 0x39, 0xc0, 0x04, 0x01, 0x98, 0x44,
+				0x38, 0xe0, 0x98, 0x10, 0x9b, 0xa8, 0x0f, 0xa8,
+			},
+		},
+	},
+	{
+		"aggregated",
+		[][]byte{
+			{
+				0xff, 0xfb, 0x14, 0x64, 0x00, 0x0f, 0xf0, 0x00,
+				0x00, 0x69, 0x00, 0x00, 0x00, 0x08, 0x00, 0x00,
+				0x0d, 0x20, 0x00, 0x00, 0x01, 0x00, 0x00, 0x01,
+				0xa4, 0x00, 0x00, 0x00, 0x20, 0x00, 0x00, 0x34,
+				0x80, 0x00, 0x00, 0x04, 0x4c, 0x41, 0x4d, 0x45,
+				0x33, 0x2e, 0x31, 0x30, 0x30, 0x55, 0x55, 0x55,
+				0x55, 0x55, 0x55, 0x55, 0x55, 0x55, 0x55, 0x55,
+				0x55, 0x55, 0x55, 0x55, 0x55, 0x55, 0x55, 0x55,
+				0x55, 0xc0, 0x65, 0xf4, 0xa0, 0x31, 0x8f, 0xce,
+				0x8d, 0x46, 0xfc, 0x8c, 0x73, 0xb9, 0x34, 0x3e,
+				0xb5, 0x03, 0x39, 0xc0, 0x04, 0x01, 0x98, 0x44,
+				0x38, 0xe0, 0x98, 0x10, 0x9b, 0xa8, 0x0f, 0xa8,
+			},
+			{
+				0xff, 0xfb, 0x14, 0x64, 0x1e, 0x0f, 0xf0, 0x00,
+				0x00, 0x69, 0x00, 0x00, 0x00, 0x08, 0x00, 0x00,
+				0x0d, 0x20, 0x00, 0x00, 0x01, 0x00, 0x00, 0x01,
+				0xa4, 0x00, 0x00, 0x00, 0x20, 0x00, 0x00, 0x34,
+				0x80, 0x00, 0x00, 0x04, 0xe6, 0x50, 0x10, 0x01,
+				0xca, 0x13, 0x94, 0x27, 0x4a, 0x4a, 0x64, 0xce,
+				0x07, 0xc2, 0x2f, 0x59, 0xc0, 0x19, 0x04, 0x05,
+				0xdf, 0xe7, 0xce, 0x65, 0x24, 0xed, 0xa4, 0xe3,
+				0xff, 0xc9, 0x00, 0x00, 0x05, 0x5f, 0x4a, 0x04,
+				0x0e, 0xc4, 0x24, 0xfd, 0x5e, 0x4a, 0x35, 0x72,
+				0x21, 0x27, 0x31, 0x08, 0x47, 0x18, 0x00, 0x06,
+				0xc4, 0x02, 0x72, 0x81, 0x89, 0xc3, 0xe4, 0x0a,
+			},
+		},
+		&rtp.Packet{
+			Header: rtp.Header{
+				Version:        2,
+				Marker:         true,
+				PayloadType:    14,
+				SequenceNumber: 17645,
+				Timestamp:      2289526357,
+				SSRC:           0x9dbb7812,
+			},
+			Payload: []byte{
+				0x00, 0x00, 0x00, 0x00,
+				0xff, 0xfb, 0x14, 0x64, 0x00, 0x0f, 0xf0, 0x00,
+				0x00, 0x69, 0x00, 0x00, 0x00, 0x08, 0x00, 0x00,
+				0x0d, 0x20, 0x00, 0x00, 0x01, 0x00, 0x00, 0x01,
+				0xa4, 0x00, 0x00, 0x00, 0x20, 0x00, 0x00, 0x34,
+				0x80, 0x00, 0x00, 0x04, 0x4c, 0x41, 0x4d, 0x45,
+				0x33, 0x2e, 0x31, 0x30, 0x30, 0x55, 0x55, 0x55,
+				0x55, 0x55, 0x55, 0x55, 0x55, 0x55, 0x55, 0x55,
+				0x55, 0x55, 0x55, 0x55, 0x55, 0x55, 0x55, 0x55,
+				0x55, 0xc0, 0x65, 0xf4, 0xa0, 0x31, 0x8f, 0xce,
+				0x8d, 0x46, 0xfc, 0x8c, 0x73, 0xb9, 0x34, 0x3e,
+				0xb5, 0x03, 0x39, 0xc0, 0x04, 0x01, 0x98, 0x44,
+				0x38, 0xe0, 0x98, 0x10, 0x9b, 0xa8, 0x0f, 0xa8,
+				0xff, 0xfb, 0x14, 0x64, 0x1e, 0x0f, 0xf0, 0x00,
+				0x00, 0x69, 0x00, 0x00, 0x00, 0x08, 0x00, 0x00,
+				0x0d, 0x20, 0x00, 0x00, 0x01, 0x00, 0x00, 0x01,
+				0xa4, 0x00, 0x00, 0x00, 0x20, 0x00, 0x00, 0x34,
+				0x80, 0x00, 0x00, 0x04, 0xe6, 0x50, 0x10, 0x01,
+				0xca, 0x13, 0x94, 0x27, 0x4a, 0x4a, 0x64, 0xce,
+				0x07, 0xc2, 0x2f, 0x59, 0xc0, 0x19, 0x04, 0x05,
+				0xdf, 0xe7, 0xce, 0x65, 0x24, 0xed, 0xa4, 0xe3,
+				0xff, 0xc9, 0x00, 0x00, 0x05, 0x5f, 0x4a, 0x04,
+				0x0e, 0xc4, 0x24, 0xfd, 0x5e, 0x4a, 0x35, 0x72,
+				0x21, 0x27, 0x31, 0x08, 0x47, 0x18, 0x00, 0x06,
+				0xc4, 0x02, 0x72, 0x81, 0x89, 0xc3, 0xe4, 0x0a,
+			},
+		},
+	},
+}
+
+func TestEncode(t *testing.T) {
+	for _, ca := range cases {
+		t.Run(ca.name, func(t *testing.T) {
+			e := &Encoder{
+				SSRC: func() *uint32 {
+					v := uint32(0x9dbb7812)
+					return &v
+				}(),
+				InitialSequenceNumber: func() *uint16 {
+					v := uint16(0x44ed)
+					return &v
+				}(),
+				InitialTimestamp: func() *uint32 {
+					v := uint32(0x88776655)
+					return &v
+				}(),
+			}
+			e.Init()
+
+			pkts, err := e.Encode(ca.frames, 0)
+			require.NoError(t, err)
+			require.Equal(t, ca.pkt, pkts[0])
+		})
+	}
+}
+
+func TestEncodeRandomInitialState(t *testing.T) {
+	e := &Encoder{}
+	e.Init()
+	require.NotEqual(t, nil, e.SSRC)
+	require.NotEqual(t, nil, e.InitialSequenceNumber)
+	require.NotEqual(t, nil, e.InitialTimestamp)
+}

pkg/formats/rtpmpeg2audio/rtpmpeg2audio.go

@@ -0,0 +1,2 @@
+// Package rtpmpeg2audio contains a RTP/MPEG2-audio decoder and encoder.
+package rtpmpeg2audio

pkg/formats/rtpmpeg2audio/testdata/fuzz/FuzzDecoder/582528ddfad69eb5

@@ -0,0 +1,2 @@
+go test fuzz v1
+[]byte("0")

pkg/formats/rtpmpeg2audio/testdata/fuzz/FuzzDecoder/b456f27416f443a7

@@ -0,0 +1,2 @@
+go test fuzz v1
+[]byte("\x00\x00\x00\x000")

pkg/formats/rtpmpeg2audio/testdata/fuzz/FuzzDecoder/d92bed38676468bc

@@ -0,0 +1,2 @@
+go test fuzz v1
+[]byte("\x00\x00000")

pkg/formats/rtpmpeg2audio/testdata/fuzz/FuzzDecoder/f34630c44c11bb13

@@ -0,0 +1,2 @@
+go test fuzz v1
+[]byte("00000")

pkg/formats/rtpmpeg4audio/decoder.go

@@ -16,6 +16,7 @@ import (
 var ErrMorePacketsNeeded = errors.New("need more packets")
 
 // Decoder is a RTP/MPEG4-audio decoder.
+// Specification: https://datatracker.ietf.org/doc/html/rfc3640
 type Decoder struct {
 	// sample rate of input packets.
 	SampleRate int
@@ -73,7 +74,7 @@ func (d *Decoder) Decode(pkt *rtp.Packet) ([][]byte, time.Duration, error) {
 	var aus [][]byte
 
 	if len(d.fragments) == 0 {
-		if pkt.Header.Marker {
+		if pkt.Marker {
 			// AUs
 			aus = make([][]byte, len(dataLens))
 			for i, dataLen := range dataLens {
@@ -117,7 +118,7 @@ func (d *Decoder) Decode(pkt *rtp.Packet) ([][]byte, time.Duration, error) {
 
 		d.fragments = append(d.fragments, payload[:dataLens[0]])
 
-		if !pkt.Header.Marker {
+		if !pkt.Marker {
 			return nil, 0, ErrMorePacketsNeeded
 		}
 

pkg/formats/rtpmpeg4audio/decoder_test.go

@@ -109,7 +109,7 @@ func TestDecodeADTS(t *testing.T) {
 	}
 }
 
-func FuzzDecoderUnmarshal(f *testing.F) {
+func FuzzDecoder(f *testing.F) {
 	d := &Decoder{
 		SampleRate:       16000,
 		SizeLength:       13,

pkg/formats/rtpmpeg4audio/encoder.go

@@ -22,6 +22,7 @@ func randUint32() uint32 {
 }
 
 // Encoder is a RTP/MPEG4-audio encoder.
+// Specification: https://datatracker.ietf.org/doc/html/rfc3640
 type Encoder struct {
 	// payload type of packets.
 	PayloadType uint8

pkg/formats/rtpvp8/decoder.go

@@ -22,6 +22,7 @@ var ErrNonStartingPacketAndNoPrevious = errors.New(
 	"received a non-starting fragment without any previous starting fragment")
 
 // Decoder is a RTP/VP8 decoder.
+// Specification: https://datatracker.ietf.org/doc/html/rfc7741
 type Decoder struct {
 	timeDecoder         *rtptime.Decoder
 	firstPacketReceived bool

pkg/formats/rtpvp8/decoder_test.go

@@ -52,7 +52,7 @@ func TestDecode(t *testing.T) {
 	}
 }
 
-func FuzzDecoderUnmarshal(f *testing.F) {
+func FuzzDecoder(f *testing.F) {
 	d := &Decoder{}
 	d.Init()
 

pkg/formats/rtpvp8/encoder.go

@@ -22,6 +22,7 @@ func randUint32() uint32 {
 }
 
 // Encoder is a RTP/VP8 encoder.
+// Specification: https://datatracker.ietf.org/doc/html/rfc7741
 type Encoder struct {
 	// payload type of packets.
 	PayloadType uint8

pkg/formats/rtpvp9/decoder.go

@@ -22,6 +22,7 @@ var ErrNonStartingPacketAndNoPrevious = errors.New(
 	"received a non-starting fragment without any previous starting fragment")
 
 // Decoder is a RTP/VP9 decoder.
+// Specification: https://datatracker.ietf.org/doc/html/draft-ietf-payload-vp9-16
 type Decoder struct {
 	timeDecoder         *rtptime.Decoder
 	firstPacketReceived bool

pkg/formats/rtpvp9/decoder_test.go

@@ -52,7 +52,7 @@ func TestDecode(t *testing.T) {
 	}
 }
 
-func FuzzDecoderUnmarshal(f *testing.F) {
+func FuzzDecoder(f *testing.F) {
 	d := &Decoder{}
 	d.Init()
 

pkg/formats/rtpvp9/encoder.go

@@ -22,6 +22,7 @@ func randUint32() uint32 {
 }
 
 // Encoder is a RTP/VP9 encoder.
+// Specification: https://datatracker.ietf.org/doc/html/draft-ietf-payload-vp9-16
 type Encoder struct {
 	// payload type of packets.
 	PayloadType uint8