predictive coding

50
• Predictive Coding https://store.theartofservice.com/the-predictive-coding- toolkit.html

Upload: vanessa-wilson

Post on 27-Dec-2015

236 views

Category:

Documents


0 download

TRANSCRIPT

• Predictive Coding

https://store.theartofservice.com/the-predictive-coding-toolkit.html

Data compression -

1 Other types of lossy compressors, such as the linear predictive coding (LPC) used with speech, are source-

based coders

https://store.theartofservice.com/the-predictive-coding-toolkit.html

GSM - Voice codecs

1 These used a system based on linear predictive

coding (LPC)

https://store.theartofservice.com/the-predictive-coding-toolkit.html

MP3 - Development

1 Manfred Schroeder was already a well-known and revered figure in the worldwide community of acoustical

and electrical engineers, but his paper was not much noticed, since it described negative results due to the particular nature of speech and the linear predictive coding (LPC) gain

present in speech

https://store.theartofservice.com/the-predictive-coding-toolkit.html

Perception - Effect of motivation and expectation

1 Instead, our brains use what he calls Predictive

coding

https://store.theartofservice.com/the-predictive-coding-toolkit.html

Bayesian approaches to brain function - Predictive coding

1 Predictive coding is a neurobiologically plausible scheme for inferring the causes of

sensory input based on minimizing prediction error.Rao RPN, Ballard DH.

Predictive coding in the visual cortex: a functional interpretation of some extra-classical receptive-field effects. Nature

Neuroscience. 1999. 2:79–87 These schemes are related formally to Kalman

filtering and other Bayesian update schemes.

https://store.theartofservice.com/the-predictive-coding-toolkit.html

Bayesian approaches to brain function - Free energy

1 This can be cast (in neurobiologically plausible terms) as predictive coding or, more generally, Bayesian filtering.

https://store.theartofservice.com/the-predictive-coding-toolkit.html

Digital signal processor - History

1 In 1976, Richard Wiggins proposed the Speak Spell (toy)|Speak Spell concept to Paul Breedlove, Larry Brantingham, and Gene Frantz at Texas Instrument's Dallas research facility. Two years later in 1978

they produced the first Speak Spell, with the technological centerpiece being the

TMS5100, the industry's first digital signal processor. It also set other milestones, being the first chip to use Linear predictive coding

to perform speech synthesis.

https://store.theartofservice.com/the-predictive-coding-toolkit.html

Silk (codec) - Details

1 Skype Limited announced that SILK can use a sampling frequency of 8, 12, 16 or 24kHz and a bit rate from 6 to 40 kbit/s. It can also use a low algorithmic latency (audio)|delay of 25 ms (20 ms frame size + 5 ms look-ahead). The reference implementation is

written in the C programming language. The codec technology is based on linear

predictive coding (LPC). The SILK executable|binary Software development

kit|SDK is available.https://store.theartofservice.com/the-predictive-coding-toolkit.html

MBONE - Mbone tools

1 * Audioconferencing: vat -t ttl destination-host/port (supports:

Linear predictive coding|LPC, PCMU, DVI4, GSM)

https://store.theartofservice.com/the-predictive-coding-toolkit.html

Texas Instruments - First speech synthesis chip

1 This resulted in the development the TMC0280 one-chip Linear predictive coding (LPC) speech

synthesizer which was the first time a single silicon chip had electronically replicated the

human voice.[http://americanhistory.si.edu/archives/spee

chsynthesis/ss_ti.htm “Smithsonian Speech Synthesis History Project”], accessed September 7, 2008 This was used in several TI commercial

products beginning with Speak Spell (toy)|Speak Spell which was introduced at the Summer Consumer Electronics Show in June 1978

https://store.theartofservice.com/the-predictive-coding-toolkit.html

Secure voice - Digital Methods using Voice Compression: MELP or MELPe

1 As part of NATO testing for new NATO standard, MELPe was tested against other candidates such as France's

HSX (Harmonic Stochastic eXcitation) and Turkey's SB-LPC (Split-Band

Linear Predictive Coding), as well as the old secure voice standards such

as FS1015 LPC-10e (2.4 kbit/s), FS1016 CELP (4.8 kbit/s) and CVSD

(16 kbit/s)https://store.theartofservice.com/the-predictive-coding-toolkit.html

Speech synthesis - Diphone synthesis

1 At runtime, the target prosody (linguistics)|prosody of a sentence is superimposed on these minimal units by means of digital signal processing techniques such as linear predictive

coding, PSOLA or MBROLA.T

https://store.theartofservice.com/the-predictive-coding-toolkit.html

Manfred R. Schroeder

1 He joined the technical staff at Bell Labs in New Jersey (1954-)

researching speech and graphics, securing forty-five patents. With Bishnu Atal he was a promotor of

linear predictive coding (1967). Still affiliated with Bell, he rejoined

University of Göttingen as Universitätsprofessor Physik (1969)

becoming professor emeritus (1991).https://store.theartofservice.com/the-predictive-coding-toolkit.html

Multimedia Information Retrieval - Feature Extraction Methods

1 Typical methods include Linear Predictive Coding in the

audio/biosignal domain,HG Kim , N Moreau, T Sikora

https://store.theartofservice.com/the-predictive-coding-toolkit.html

Video compression - Coding methods

1 Other types of lossy compressors, such as the linear predictive coding (LPC) used with speech, are source-

based coders

https://store.theartofservice.com/the-predictive-coding-toolkit.html

Opus (audio format)

1 Opus can be adjusted seamlessly between high and low bit rate|

bitrates, and internally, it transitions between linear predictive coding at lower bitrates and transform coding

at higher bitrates (as well as a hybrid for a short overlap)

https://store.theartofservice.com/the-predictive-coding-toolkit.html

Opus (audio format) - Features

1 The SILK layer that specializes in speech signals is based on linear predictive coding (LPC) and an

optional Long-Term Prediction filter

https://store.theartofservice.com/the-predictive-coding-toolkit.html

High Efficiency Video Coding - Range extensions

1 **Residual DPCM (RDPCM), allowing a vertical or horizontal spatial-

predictive coding of residual data in transform skip and transform-

quantization bypass blocks (which can be selected for use in intra blocks, inter blocks, or both).

https://store.theartofservice.com/the-predictive-coding-toolkit.html

Warped linear predictive coding

1 'Warped linear predictive coding' ('warped LPC' or 'WLPC') is a variant of linear predictive coding in which the

spectral representation of the system is modified, for example by replacing the unit delays used in an LPC implementation with

first-order allpass filters. This can have advantages in reducing the bitrate required

for a given level of perceived audio quality/intelligibility, especially in wideband

audio coding.https://store.theartofservice.com/the-predictive-coding-toolkit.html

Algebraic code-excited linear prediction

1 'Algebraic code-excited linear prediction' ('ACELP') is a patentedUS

Patent 5717825 Algebraic code-excited linear prediction speech coding method speech coding

algorithm by VoiceAge | VoiceAge Corporation in which a limited set of pulses is distributed as excitation to

a Linear predictive coding|linear prediction filter.

https://store.theartofservice.com/the-predictive-coding-toolkit.html

Code-excited linear prediction

1 At the time, it provided significantly better quality than existing low bit-rate algorithms, such as residual-

excited linear prediction and linear predictive coding vocoders (e.g., FS-

1015)

https://store.theartofservice.com/the-predictive-coding-toolkit.html

ILBC

1 The algorithm is a version of block-independent linear predictive coding, with the choice of data frame lengths

of 20 and 30 milliseconds

https://store.theartofservice.com/the-predictive-coding-toolkit.html

Network Voice Protocol

1 NVP was used to send speech between distributed sites on the ARPANET using several different voice-encoding techniques, including linear predictive coding (LPC) and continuously

variable slope delta modulation (CVSD). Cooperating researchers included Steve Casner, Randy Cole, and Paul Raveling (ISI); Jim Forgie (Lincoln Laboratory); Mike McCammon (Culler-

Harrison); John Markel (Speech Communications Research Laboratory); and John Makhoul (Bolt,

Beranek and Newman).

https://store.theartofservice.com/the-predictive-coding-toolkit.html

Line spectral pairs

1 'Line spectral pairs' ('LSP') or 'line spectral frequencies' ('LSF') are used to represent linear predictive coding|linear prediction coefficients

(LPC) for transmission over a channel

https://store.theartofservice.com/the-predictive-coding-toolkit.html

Secure telephone - Historically significant products

1 In 1989 an Irish company called Intrepid developed one of the most

advanced secure phones. Called Milcode, the phone was the first to

implement Code Excited Linear Prediction (or CELP) which

dramatically improved voice quality and user operability over previous LPC (Linear Predictive Coding) and

LPC-10e versions.https://store.theartofservice.com/the-predictive-coding-toolkit.html

Log area ratio

1 'Log area ratios' ('LAR') can be used to represent reflection

coefficients (another form for Linear predictive coding|linear prediction

coefficients) for transmission over a channel. While not as efficient as line spectral pairs (LSPs), log area ratios

are much simpler to compute. Let r_k be the kth reflection coefficient of a

filter, the kth LAR is:https://store.theartofservice.com/the-predictive-coding-toolkit.html

Golomb coding - Applications

1 Several lossless audio data compression|audio codecs, such as Shorten (file format)|Shorten,[http://www.etree.org/shnutils/short

en/support/doc/shorten.txt man shorten] FLAC,[http://flac.sourceforge.net/document

ation_format_overview.html FLAC documentation: format overview] Apple

Lossless, and MPEG-4 ALS, use a Rice code after the Linear predictive coding|linear

prediction step (called adaptive FIR filter in Apple Lossless).

https://store.theartofservice.com/the-predictive-coding-toolkit.html

Linear predictive coding

1 'Linear predictive coding' ('LPC') is a tool used mostly in audio signal

processing and speech processing for representing the spectral envelope of

a Digital data|digital signal (information theory)|signal of Speech

communication|speech in data compression|compressed form, using the information of a linear prediction|

linear predictive modelhttps://store.theartofservice.com/the-predictive-coding-toolkit.html

Linear predictive coding - Applications

1 Paul Lansky made the well-known computer music piece

notjustmoreidlechatter using linear predictive

coding.[http://www.music.princeton.edu/~paul/liner_notes/morethanidlech

atter.html]

https://store.theartofservice.com/the-predictive-coding-toolkit.html

Release time -

1 'Analysis/resynthesis' is a form of synthesis that uses a series of

bandpass filters or Fourier transforms to analyze the harmonic content of a sound. The resulting analysis data is

then used in a second stage to resynthesize the sound using a band

of oscillators. The vocoder, linear predictive coding, and some forms of

speech synthesis are based on analysis/resynthesis.

https://store.theartofservice.com/the-predictive-coding-toolkit.html

Codec2 - Overview

1 The LSP coefficients represent the Linear predictive coding|Linear

Predictive Coding (LPC) model in the frequency domain, and lend

themselves to

https://store.theartofservice.com/the-predictive-coding-toolkit.html

G.728

1 The linear prediction is calculated backwards with a 50th order linear predictive coding

filter

https://store.theartofservice.com/the-predictive-coding-toolkit.html

Lossless JPEG - Lossless mode of operation

1 Unlike the lossy mode which is based on the DCT, the lossless coding

process employs a simple predictive coding model called differential pulse

code modulation (DPCM)

https://store.theartofservice.com/the-predictive-coding-toolkit.html

Harmonic Vector Excitation Coding - Linear Predictive Coding

1 HVXC uses Linear predictive coding (LPC) with block-wise adaptation every 20ms. The LPC parameters are transformed to Line spectral

pairs|Line spectral pair (LSP) coefficients, which are jointly quantized. The LPC residual signal is classified as either voiced or unvoiced. In the

case of voiced speech, the residual is coded in a parametric representation (operating as a

vocoder), while in the case of unvoiced speech, the residual waveform is quantized (thus

operating as hybrid speech codec).

https://store.theartofservice.com/the-predictive-coding-toolkit.html

Audio Lossless Coding

1 The short-term predictor is similar to Free Lossless Audio Codec|FLAC in its

operation - it is a quantized Linear predictive coding|LPC predictor with

a losslessly coded residual using Golomb coding|Golomb Rice Coding

or Block Gilbert Moore Coding (BGMC)

https://store.theartofservice.com/the-predictive-coding-toolkit.html

TI-99/4A - Features

1 The synthesizer used a variant of linear predictive coding and had a small in-built

vocabulary

https://store.theartofservice.com/the-predictive-coding-toolkit.html

Adaptive Multi-Rate audio codec - Features

1 ** Linear predictive coding (LPC) is used to synthesize the speech from a

residual waveform. The LPC parameters are encoded as Line

spectral pairs (LSP).

https://store.theartofservice.com/the-predictive-coding-toolkit.html

Paul Lansky - Biography

1 Following that came More Than Idle Chatter, the six compositions of

which focus on processings of the human voice using linear predictive

coding, granular synthesis, and plucked string synthesis; its three highlights are Granular synthesis|granular synth pieces called Idle

Chatter, Just_more_idle_chatter, and Notjustmoreidlechatter, which look at

the same thing from multiple perspectives

https://store.theartofservice.com/the-predictive-coding-toolkit.html

FS-1015

1 The standard was finished 1984. The algorithm used is a linear predictive coding vocoder. The vocoder enables

understandable speech, but the quality is very unnatural and synthetic. File size is 20 times

smaller than MP3 - it is very small.

https://store.theartofservice.com/the-predictive-coding-toolkit.html

System 16 - G80 specifications

1 *** Speech synthesis|Speech synthesizer: General Instrument

SP0256|General Instrument SP0250 (linear predictive coding)

https://store.theartofservice.com/the-predictive-coding-toolkit.html

Audio timescale-pitch modification - Pitch scaling

1 A process that preserves the formants and character of a voice

involves analyzing the signal with a vocoder|channel vocoder or Linear predictive coding|LPC vocoder plus

any of several pitch detection algorithms and then resynthesizing it at a different fundamental frequency.

https://store.theartofservice.com/the-predictive-coding-toolkit.html

MPEG-2 - Patents (U.S. only)

1 | 5291284 || 23 jul 1991 || 01 mar 1994 || 12 dec 1989 || 01 mar 2011 || Predictive coding and decoding with

error drift reduction || http://patft1.uspto.gov/netacgi/nph-

Parser?patentnumber=5291284 file+20: [2011, 7, 23] pct_file+20:

[2009, 12, 12] grant+17:[2011, 3, 1] || British Telecommunications plc

https://store.theartofservice.com/the-predictive-coding-toolkit.html

MPEG-2 - Patents (U.S. only)

1 | 5317397 || 29 May 1992 || 31 May 1994 || 29 May 1992 || 29 May 2012

|| Predictive coding using spatial-temporal filtering and plural motion

vectors || http://patft1.uspto.gov/netacgi/nph-

Parser?patentnumber=5317397 file+20: [2012, 5, 29] grant+17:

[2011, 5, 31] || Toshiba

https://store.theartofservice.com/the-predictive-coding-toolkit.html

MPEG-2 - Patents (U.S. only)

1 | 5103307 || 18 jan 1991 || 07 apr 1992 || 18 jan 1991 || 18 jan 2011 ||

Interframe predictive coding/decoding system for varying

interval between independent frames ||

http://patft1.uspto.gov/netacgi/nph-Parser?patentnumber=5103307 file+20: [2011, 1, 18] grant+17:

[2009, 4, 7] || JVC|Victor Companyhttps://store.theartofservice.com/the-predictive-coding-toolkit.html

Vocoder - Modern implementations

1 * LPC-10, Federal Information Processing Standard|FIPS Pub 137,

2400 bit/s, which uses linear predictive coding

https://store.theartofservice.com/the-predictive-coding-toolkit.html

Echo 2

1 LPC (linear predictive coding) was the speech synthesis technology

used, which allowed applications to encode speech data in a compact form. The Echo II used the Texas

Instruments LPC Speech Chips|TMS 5220 LPC Speech Chip which was

popular in other speech synthesizers

https://store.theartofservice.com/the-predictive-coding-toolkit.html

Texas Instruments LPC Speech Chips - Theory

1 Speech communication|Speech data is stored through pitch-excited linear

predictive coding (PE-LPC), where words are created by a lattice filter, selectably fed by

either an excitation Read-only memory|ROM (containing a glottal pulse waveform) or an LFSR (linear feedback shift register) noise

generator. Linear predictive coding achieves a vast reduction in data volume needed to

recreate intelligible speech data.

https://store.theartofservice.com/the-predictive-coding-toolkit.html

Texas Instruments LPC Speech Chips - History

1 The TMC0280/TMS5100 was the first self-contained linear predictive coding|LPC speech synthesizer Integrated Circuit|IC ever made

https://store.theartofservice.com/the-predictive-coding-toolkit.html