Commits

Breadcrumbs

History for

on527ef09

User selector

All users

Datepicker

All time

Commit History

Commits on Nov 24, 2015

• fix bug with reporting dropped hash count for an empty database

  dpwe

  committed
  dd0b2b9

Commits on Oct 31, 2015

• Removed ID+1 fudge; now removing the very first item in the dbase is slow, but it's compatible.

  dpwe

  committed
  a887099

• Provide remove and list commands; IDs in table are now +1

  dpwe

  committed
  07f7641

Commits on Dec 22, 2014

• added newmerge command; fixed hash_table to merge hash tables of different bucketsize

  dpwe

  committed
  0303b9c

Commits on Nov 1, 2014

• refactored and streamlined match; store maxtime as maxtimebits

  dpwe

  committed
  88add71

Commits on Oct 29, 2014

• Print the number and percentage of dropped hashes when saving dbase

  dpwe

  committed
  1633632

Commits on Oct 26, 2014

• fix problem adding to hash_table; further optimization of settings for dpwe_matcher; default is 8 cores

  dpwe

  committed
  9c7afe1

• additional cleanup; config improved for uspop+gtzan

  dpwe

  committed
  4372b8a

• sort hits by length-normalized raw counts

  dpwe

  committed
  da4d979

• fixed minor bug in multiproc add/merge, added ht.reset()

  dpwe

  committed
  cc7f132

• avoid realloc/copy in get_hits - 20x faster for large queries

  dpwe

  committed
  08652e5

• ensure get_hits returns int32; avoid equal.outers in match_hashes - saves 75% of the time, and avoids exploding memory usage

  dpwe

  committed
  2a88b11

Commits on Oct 25, 2014

• hash table hits are now handled as np arrays; added support for search depth

  dpwe

  committed
  25f00b9

Commits on Sep 20, 2014

• further surface changes to please pylint across all source files

  dpwe

  committed
  aa5254e

• Many changes and refactoring to please pylint; 'make test' added

  dpwe

  committed
  07609e4

Commits on Sep 6, 2014

• final? fix to ht.merge

  dpwe

  committed
  e91e672

• fix to randperm on ht.merge; first cut at joblib parallelization of match

  dpwe

  committed
  323c4d8

• alternate multiprocessing runs n parallel jobs for create

  dpwe

  committed
  55030e3

• efforts to streamline findepeaks fwd pass; added merge command to merge hash tables

  dpwe

  committed
  efd0972

Commits on Sep 3, 2014

• more optimizations - streamlining hash_table:store loop, early quit in peaks2landmarks loop

  dpwe

  committed
  9883c9d

• optimizations - sorted by hashes in hash_table:store, avoid call to spreadpeaks within find_landmarks

  dpwe

  committed
  b22672a

Commits on Aug 1, 2014

• better messages; added precompute option

  dpwe

  committed
  77961c0

Commits on May 29, 2014

• added --samplerate option, now also read back from hashtable

  dpwe

  committed
  c62b0a9

Commits on May 27, 2014

• now respects --density; reports NO MATCH if matching below threshold

  dpwe

  committed
  ecaf0b3

• fixed bug and improved list comprehensions for hash table querying

  dpwe

  committed
  ea8e9b8

• CLI added to audfprint.py, nicer messages

  dpwe

  committed
  843420e

• Now reads databases written by Matlab audfprint

  dpwe

  committed
  0f082f7

Commits on May 26, 2014

• implemented audfprint_match, see dotest block therein

  dpwe

  committed
  eefa119

• First version of python port of audfprint; hashes agree with matlab (modulo origin-1 timebase), hash table loads and saves

  dpwe

  committed
  5194126

Loading