CS Spring 2011 CS 414 – Multimedia Systems Design Lecture 11 – MP3 Audio & Introduction to MPEG-4 (Part 6) Klara Nahrstedt Spring 2011

CS Spring 2011 Administrative MP1 – deadline February 18

Outline MP3 Audio Encoding MPEG-2 and Intro to MPEG-4 Reading:  Media Coding book, Section –  Recommended Paper on MP3: Davis Pan, “A Tutorial on MPEG/Audio Compression”, IEEE Multimedia, pp. 6-74, 1995  Recommended books on JPEG/ MPEG Audio/Video Fundamentals: Haskell, Puri, Netravali, “Digital Video: An Introduction to MPEG-2”, Chapman and Hall, 1996 CS Spring 2011

MPEG-1 Audio Encoding Characteristics  Precision 16 bits  Sampling frequency: 32KHz, 44.1 KHz, 48 KHz  3 compression layers: Layer 1, Layer 2, Layer 3 (MP3) Layer 3: kbps, target 64 kbps Layer 2: kbps, target 128 kbps Layer 1: kbps, target 192 kbps CS Spring 2011

MPEG Audio Encoding Steps CS Spring 2011

MPEG Audio Filter Bank Filter bank divides input into multiple sub-bands (32 equal frequency sub-bands) Sub-band i defined - filter output sample for sub-band i at time t, C[n] – one of 512 coefficients, x[n] – audio input sample from 512 sample buffer CS Spring 2011

MPEG Audio Psycho-acoustic Model MPEG audio compresses by removing acoustically irrelevant parts of audio signals Takes advantage of human auditory systems inability to hear quantization noise under auditory masking Auditory masking: occurs when ever the presence of a strong audio signal makes a temporal or spectral neighborhood of weaker audio signals imperceptible. CS Spring 2011

MPEG/audio divides audio signal into frequency sub-bands that approximate critical bands. Then we quantize each sub-band according to the audibility of quantization noise within the band

MPEG Audio Bit Allocation This process determines number of code bits allocated to each sub-band based on information from the psycho- acoustic model Algorithm: 1. Compute mask-to-noise ratio: MNR=SNR-SMR Standard provides tables that give estimates for SNR resulting from quantizing to a given number of quantizer levels 2. Get MNR for each sub-band 3. Search for sub-band with the lowest MNR 4. Allocate code bits to this sub-band. If sub-band gets allocated more code bits than appropriate, look up new estimate of SNR and repeat step 1 CS Spring 2011

MP3 Audio Format CS Spring 2011 Source:

MPEG Audio Comments Precision of 16 bits per sample is needed to get good SNR ratio Noise we are getting is quantization noise from the digitization process For each added bit, we get 6dB better SNR ratio Masking effect means that we can raise the noise floor around a strong sound because the noise will be masked away Raising noise floor is the same as using less bits and using less bits is the same as compression CS Spring 2011

MPEG-2 Standard Extension MPEG-1 was optimized for CD-ROM and apps for 1.5 Mbps (video strictly non- interleaved) MPEG-2 adds to MPEG-1:  More aspect ratios: 4:2:2, 4:4:4  Progressive and interlaced frame coding  Four scalable modes spatial scalability, data partitioning, SNR scalability, temporal scalability CS Spring 2011

MPEG-1/MPEG-2 Pixel-based representations of content  takes place at the encoder Lack support for content manipulation  e.g., remove a date stamp from a video  turn off “current score” visual in a live game Need support manipulation and interaction if the video is aware of its own content CS Spring 2011

MPEG-4 Compression CS Spring 2011

Interact With Visual Content CS Spring 2011

Original MPEG-4 Conceived in 1992 to address very low bit rate audio and video (64 Kbps) Required quantum leaps in compression  beyond statistical- and DCT-based techniques  committee felt it was possible within 5 years Quantum leap did not happen CS Spring 2011

The “New” MPEG-4 Support object-based features for content Enable dynamic rendering of content  defer composition until decoding Support convergence among digital video, synthetic environments, and the Internet CS Spring 2011

MPEG-4 Components Systems – defines architecture, multiplexing structure, syntax Video – defines video coding algorithms for animation of synthetic and natural hybrid video (Synthetic/Natural Hybrid Coding) Audio – defines audio/speech coding, Synthetic/Natural Hybrid Coding such as MIDI and text-to-speech synthesis integration Conformance Testing – defines compliance requirements for MPEG-4 bitstream and decoders Technical report DSM-CC Multimedia Integration Framework – defines Digital Storage Media – Command&Control Multimedia Integration Format; specifies merging of broadcast, interactive and conversational multimedia for set-top-boxes and mobile stations CS Spring 2011

MPEG-4 Example ISO N3536 MPEG4 CS Spring 2011

MPEG-4 Example ISO N3536 MPEG4 CS Spring 2011

MPEG-4 Example Daras, P. MPEG-4 Authoring Tool, J. Applied Signal Processing, 9, 1-18, 2003 CS Spring 2011

Interactive Drama CS Spring 2011

MPEG-4 Characteristics and Applications CS Spring 2011

Media Objects An object is called a media object  real and synthetic images; analog and synthetic audio; animated faces; interaction Compose media objects into a hierarchical representation  form compound, dynamic scenes CS Spring 2011

Composition Scene furniture deskglobe character voicesprite ISO N3536 MPEG4 CS Spring 2011

Video Syntax Structure CS Spring 2011 New MPEG-4 Aspect: Object-based layered syntactic structure

Content-based Interactivity Achieves different qualities for different objects with a fine granularity in spatial resolution, temporal resolution and decoding complexity Needs coding of video objects with arbitrary shapes Scalability  Spatial and temporal scalability  Need more than one layer of information (base and enhancement layers) CS Spring 2011

Examples of Base and Enhancement Layers CS Spring 2011

Coding of Objects Each VOP corresponds to an entity that after being coded is added to the bit stream Encoder sends together with VOP  Composition information where and when each VOP is to be displayed Users are allowed to change the composition of the entire scene displayed by interacting with the composition information CS Spring 2011

Spatial Scalability CS Spring 2011 VOP which is temporally coincident with I-VOP in the base layer, is encoded as P-VOP in the enhancement layer. VOP which is temporally coincident with P-VOP in the base layer is encoded as B-VOP in the enhancement layer.

Temporal Scalability CS Spring 2011

Composition (cont.) Encode objects in separate channels  encode using most efficient mechanism  transmit each object in a separate stream Composition takes place at the decoder, rather than at the encoder  requires a binary scene description (BIFS) BIFS is low-level language for describing:  hierarchical, spatial, and temporal relations CS Spring 2011

MPEG-4 Rendering ISO N3536 MPEG4 CS Spring 2011

Interaction as Objects Change colors of objects Toggle visibility of objects Navigate to different content sections Select from multiple camera views  change current camera angle Standardizes content and interaction  e.g., broadcast HDTV and stored DVD CS Spring 2011

Hierarchical Model Each MPEG-4 movie composed of tracks  each track composed of media elements (one reserved for BIFS information)  each media element is an object  each object is a audio, video, sprite, etc. Each object specifies its:  spatial information relative to a parent  temporal information relative to global timeline CS Spring 2011

Synchronization Global timeline (high-resolution units)  e.g., 600 units/sec Each continuous track specifies relation  e.g., if a video is 30 fps, then a frame should be displayed every 33 ms. Others specify start/end time CS Spring 2011

MPEG-4 parts MPEG-4 part 2  Includes Advanced Simple Profile, used by codecs such as Quicktime 6 MPEG-4 part 10  MPEG-4 AVC/H.264 also called Advanced Video Coding  Used by coders such as Quicktime 7  Used by high-definition video media like Blu- ray Disc CS Spring 2011