acknowledgements for ansi/nist-itl 1-2011 - NIST Visual Image ...

acknowledgements for ansi/nist-itl 1-2011 - NIST Visual Image ... acknowledgements for ansi/nist-itl 1-2011 - NIST Visual Image ...

biometrics.nist.gov
from biometrics.nist.gov More from this publisher
11.07.2015 Views

ANSI/NIST-ITL 1-2011 - UPDATE 2013 DRAFT VERSION• This field should be present if Field 11.013: Container / CONT is notpresent in the record.The information items in the field are described below.oThe first information item (codec code / CODC) is mandatory if this field isused and indicates the single codec type used for all audio segments in therecord. This standard does not accommodate multiple codec types within asingle record. CODC shall be a numeric entry selected from the Codec codecolumn of Table 75 Codec Type Codes.If the codec code is identified as Other (CODC=2), the final information item(comments / COM) shall be used to describe the codec.See http://www.nist.gov/itl/iad/ig/ansi_standard.cfm for additional codecs.Such additional codecs (not listed in Table 75 Codec Type Codes) shall use thevalue CODC=1 and the 'reference code' shall be entered in external codecreference code / ECOD.Table 75 Codec Type CodesCodec typeCodec codeLinear PCM 0Codec type reference 1Other 2Floating-point linear PCM 3ITU-T G.711 137 (PCM): μ-law with4forward order digital samplesITU-T G.711 (PCM): μ-law with5reverse order digital samplesITU-T G.711 (PCM): A-law with6forward order digital samplesITU-T G.711 (PCM): A-law with7reverse order digital samplesooThe second information item (sampling rate number / SRTN) is mandatoryand indicates the number of digital samples per second that represent a secondof analog voice data upon conversion to an acoustic signal. The sampling rate isexpressed in Hz and must be an integer value. Acceptable values are between 1and 100,000,000 Hz, but unknown or variable sampling rates shall be given thevalue of 0. Common values of SRTN are 8000, 11025, 16000, 22050, 32000,44100, and 48000 Hz. The value of 0 shall only be used to indicate unknown orvariable sampling rate.The third information item (bit depth count / BITD) is mandatory andindicates the number of bits that are used to represent a single digital sample ofvoice data. Acceptable values are between 1 and 64, inclusive. Encoders ofunknown or variable bit depth shall be given the value of 0. (This field is notintended to be an indication of the actual dynamic range of the voice data.)137ITU-T G.711, Pulse code modulation (PCM) of voice frequencies, is available athttp://www.itu.int/rec/T-REC-G.711/eMay, 2013 DRAFT VERSION UPDATE 2013 Page 249

ANSI/NIST-ITL 1-2011 - UPDATE 2013 DRAFT VERSIONoooooChanges to the bit depth should be logged in Field 98.900: Audit log / ALF orField 11.902: Annotation information / ANN audit logs. Common values forBIT are 8, 16, 24, and 32 bits.The fourth information item (endian code / ENDC) is mandatory and indicateswhich byte goes first for digital samples containing two or more bytes. Thevalues for ENDC are 0=big, 1=little, or 2=native endian. (ENDC is optional andignored for digital samples that do not contain two or more integer multiples ofbytes.)The fifth information item (fixed point indicator / FPNT) is mandatory andindicates the digital sample representation. The value is 0 if the digital samplesare represented as fixed-point or 1 if the samples are floating-point.The sixth information item (channel count / CHC) is mandatory and gives theinteger number of channels of data represented in the digital voice data file. Thenumber of channels must be between 1 and 99, inclusive. Common values forCHC are 1 and 2 channels.The seventh information item (external codec reference code / ECOD) onlyappears if CODC = 1. It contains the 'reference code' from the table of codecsavailable at http://www.nist.gov/itl/iad/ig/ansi_standard.cfm.The eighth information item (comments / COM) is an optional, unrestrictedtext string of up to 4000 characters in length. It can contain additionalinformation about the codec or additional instructions for reconstruction ofaudio output from the stored digital data. Codec parameters shall be specified inthis information item when required for unambiguous decoding. This itemshould include a description of any noise reduction processing or equalizationthat must be applied to faithfully render the voice recording.8.11.15 Field 11.015: Premliminary signal quality / PSQoooThis field is optional and gives an assessment of the general “quality” of thevoice recording. There may be as many as 9 PSQ subfields for the audio fileto indicate different types of quality assessments.The first information item (quality value / QVU) is mandatory if this field isused and shall indicate the general quality as an integer value between 0 (lowquality) and 100 (high quality). A value of 255 indicates that quality was notassessed.A second information item, an algorithm vendor identification / QAV, ismandatory if this field is used and shall specify the ID of the vendor of thequality assessment algorithm used to calculate the quality score. This 4-digithex value (See Section 5.5 Character types ) is assigned by IBIA andexpressed as four characters. The IBIA maintains the Vendor Registry ofCBEFF Biometric Organizations that map the value in this field to a registeredorganization. For algorithms not registered with the IBIA, the value of 0x00shall be used.A third information item, algorithm product identification / QAP, is250

ANSI/<strong>NIST</strong>-ITL 1-<strong>2011</strong> - UPDATE 2013 DRAFT VERSIONoooooChanges to the bit depth should be logged in Field 98.900: Audit log / ALF orField 11.902: Annotation in<strong>for</strong>mation / ANN audit logs. Common values <strong>for</strong>BIT are 8, 16, 24, and 32 bits.The fourth in<strong>for</strong>mation item (endian code / ENDC) is mandatory and indicateswhich byte goes first <strong>for</strong> digital samples containing two or more bytes. Thevalues <strong>for</strong> ENDC are 0=big, 1=little, or 2=native endian. (ENDC is optional andignored <strong>for</strong> digital samples that do not contain two or more integer multiples ofbytes.)The fifth in<strong>for</strong>mation item (fixed point indicator / FPNT) is mandatory andindicates the digital sample representation. The value is 0 if the digital samplesare represented as fixed-point or 1 if the samples are floating-point.The sixth in<strong>for</strong>mation item (channel count / CHC) is mandatory and gives theinteger number of channels of data represented in the digital voice data file. Thenumber of channels must be between 1 and 99, inclusive. Common values <strong>for</strong>CHC are 1 and 2 channels.The seventh in<strong>for</strong>mation item (external codec reference code / ECOD) onlyappears if CODC = 1. It contains the 'reference code' from the table of codecsavailable at http://www.<strong>nist</strong>.gov/<strong>itl</strong>/iad/ig/<strong>ansi</strong>_standard.cfm.The eighth in<strong>for</strong>mation item (comments / COM) is an optional, unrestrictedtext string of up to 4000 characters in length. It can contain additionalin<strong>for</strong>mation about the codec or additional instructions <strong>for</strong> reconstruction ofaudio output from the stored digital data. Codec parameters shall be specified inthis in<strong>for</strong>mation item when required <strong>for</strong> unambiguous decoding. This itemshould include a description of any noise reduction processing or equalizationthat must be applied to faithfully render the voice recording.8.11.15 Field 11.015: Premliminary signal quality / PSQoooThis field is optional and gives an assessment of the general “quality” of thevoice recording. There may be as many as 9 PSQ subfields <strong>for</strong> the audio fileto indicate different types of quality assessments.The first in<strong>for</strong>mation item (quality value / QVU) is mandatory if this field isused and shall indicate the general quality as an integer value between 0 (lowquality) and 100 (high quality). A value of 255 indicates that quality was notassessed.A second in<strong>for</strong>mation item, an algorithm vendor identification / QAV, ismandatory if this field is used and shall specify the ID of the vendor of thequality assessment algorithm used to calculate the quality score. This 4-digithex value (See Section 5.5 Character types ) is assigned by IBIA andexpressed as four characters. The IBIA maintains the Vendor Registry ofCBEFF Biometric Organizations that map the value in this field to a registeredorganization. For algorithms not registered with the IBIA, the value of 0x00shall be used.A third in<strong>for</strong>mation item, algorithm product identification / QAP, is250

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!