1. 27 Jan, 2024 7 commits
    • 郭飞's avatar
      typo · 7fa7bd65
      郭飞 authored
      want do use -> want to use
      7fa7bd65
    • Stefan Westerfeld's avatar
      Merge branch 'multi-key-decode' · ce4fefab
      Stefan Westerfeld authored
      * multi-key-decode:
        Remove blank line in wmspeed.cc.
        BlockDecoder: use mix_or_linear_decode from ClipDecoder as well
        Change result order for same time: AB blocks should go after B block
        SyncFinder: merge init_up_down and get_sync_bits
        README.adoc: document key names and audiowmark get using multiple keys
        Refactor: implement parse_key by using parse_key_list.
        Use checks during string to number conversions to avoid invalid input.
        Set permissions on newly generated keys: only allow user to read/write key.
        WavData: reserve memory if possible before loading samples (performance).
        For results with the same time, sort by bits to get deterministic output.
        Print key names and speeds properly in default output.
        Deal with special chars that need escaping for gen-key --name.
        Generate properly escaped JSON string for key names in JSON output.
        Include speed in json output matches.
        Print out key names in json output.
        Key: support loading key names from key file.
        Support gen-key --name [ <key_name> ] option, store name in key file.
        Fix unused lambda capture warning.
        SyncFinder: use thread pool to perform fft for search_approx (performance)
        Fix audiowmark cmp output for speed detection
        More speed detection cleanups.
        Refactor/cleanup speed detection code.
        Simplify speed detection vector code using push_back instead emplace_back.
        Don't return speed 1 as search result from speed detection.
        Merge key loop into the detect speed function (performance).
        Refactor speed detection API to use a key list instead of a single key.
        Use ThreadPool in ClipDecoder for convolutional decoder (performance).
        Use ThreadPool to speed up BlockDecoder convolutional pattern decoding.
        Properly sort results after parallel sync refine.
        Use ThreadPool for refining sync results (performance).
        Add name member to Key.
        Reduce frame splitting for search_approx (performance).
        Use more threads in SyncFinder::search_approx (performance).
        Use thread pool in SyncFinder search to speed up multi-key sync.
        SyncFinder: remove sync_bits from KeyResult structure.
        Move key list loop into SyncFinder/BlockDecoder/ClipDecoder.
        Avoid global state for sync / data frame positions.
        Support more than one --key / --test-key option for audiowmark get/cmp.
        Get rid of single global key for PRNG.
      Signed-off-by: Stefan Westerfeld's avatarStefan Westerfeld <stefan@space.twc.de>
      ce4fefab
    • Stefan Westerfeld's avatar
      dfaf5de7
    • Stefan Westerfeld's avatar
    • Stefan Westerfeld's avatar
    • Stefan Westerfeld's avatar
    • Stefan Westerfeld's avatar
  2. 18 Jan, 2024 8 commits
  3. 17 Jan, 2024 9 commits
  4. 16 Jan, 2024 3 commits
  5. 15 Jan, 2024 2 commits
  6. 14 Jan, 2024 6 commits
  7. 13 Jan, 2024 3 commits
  8. 11 Jan, 2024 2 commits