/ Data / Fundamentals / Production / Tutorials / Software / Home
2.1.4 Speech Files: NIST's SPHERE Format
Section 2.1.4: NIST's SPHERE Format

Audio data distributed by the Linguistic Data Consortium (LDC) is typicaly distributed on CDROM using the NIST SPHERE format. A SPHERE file consists of a single fixed size header followed by binary audio data.

The header is organized as name value pairs in a 1024-byte blocked, ASCII structure placed at the beginning of the file. The binary data can be in either big-endian or little-endian byte order. Click here for a more detailed description of the SPHERE format.

An example of a SPHERE header is shown below. Click here or use the "Save Link As" feature in your browser to download this example.

SPHERE Header Example:

    database_id -s3 RM1
    database_version -s3 1.0
    utterance_id -s11 aks0_st0783
    channel_count -i 1
    sample_count -i 50074
    sample_rate -i 16000
    sample_min -i -2032
    sample_max -i 2708
    sample_n_bytes -i 2
    sample_byte_format -s2 01
    sample_sig_bits -i 16

    (binary audio data follows)


    (type of header)
    (size of header)
    (databse type)
    (database version)
    (utterance identifier)
    (number of channels)
    (number of samples)
    (sample rate)
    (minimum amplitude)
    (maximum amplitude)
    (number of bytes per sample)
    (byte format)
    (number of bits per sample)
    (ends the header)

    (signed short integer-valued data)
Section 2.1.4: NIST's SPHERE Format
Table of Contents   Section Contents   Previous Page Up Next Page
      Glossary / Help / Support / Site Map / Contact Us / ISIP Home