US 12,380,902 B2
Vector quantizer correction for audio codec system
Marcin Ciolek, Gdansk (PL); Michal Sulewski, Gdańsk (PL); Raul A. Casas, Doylestown, PA (US); Samer Lutfi Hijazi, San Jose, CA (US); and Mihailo Kolundzija, Lausanne (CH)
Assigned to CISCO TECHNOLOGY, INC., San Jose, CA (US)
Filed by Cisco Technology, Inc., San Jose, CA (US)
Filed on Dec. 14, 2023, as Appl. No. 18/540,060.
Claims priority of provisional application 63/591,249, filed on Oct. 18, 2023.
Prior Publication US 2025/0131931 A1, Apr. 24, 2025
Int. Cl. G10L 19/038 (2013.01); G10L 19/00 (2013.01); G10L 19/005 (2013.01)
CPC G10L 19/038 (2013.01) [G10L 19/00 (2013.01); G10L 19/005 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A method comprising:
vector quantizing input vectors representative of audio into an original sequence including original indices of codewords of a codebook;
generating candidate sequences including indices of the codewords and that have respective starting transitions from an initial index of the original sequence to respective ones of all possible indices, wherein generating includes evaluating, for each candidate sequence, transition costs for transitions between the indices based on linear combinations of (i) transition probabilities of the transitions, and (ii) distances between the codewords represented by the indices and the input vectors that correspond to the indices;
summing the transition costs for each candidate sequence into a total transition cost, to produce total transition costs for corresponding ones of the candidate sequences;
selecting a preferred candidate sequence of the candidate sequences that has a lowest total transition cost; and
transmitting or storing the preferred candidate sequence in place of the original sequence.