gurushida/mnemophonix

A simple audio fingerprinting system

mnemophonix's Issues

MKV audio extract fail with Segfault

Hello, I need to deduplicate video file with many codec like AC3 5.1 / EAC3 5.1 and more than one audio track.
I think I need to write the ffmpeg part myself to downscale and extract all audio stream as a pcm_s16le ?

Else it always end with segfault :

...
    Stream #0:0(fre): Audio: pcm_s16le ([1][0][0][0] / 0x0001), 44100 Hz, 5.1(side), s16, 4233 kb/s (default)
    Metadata:
      title           : Français VFI AC3 5.1 @ 448 Kbps
      BPS             : 448000
      BPS-eng         : 448000
      DURATION        : 02:28:51.680000000
      DURATION-eng    : 02:28:51.680000000
      NUMBER_OF_FRAMES: 279115
      NUMBER_OF_FRAMES-eng: 279115
      NUMBER_OF_BYTES : 500174080
      NUMBER_OF_BYTES-eng: 500174080
      _STATISTICS_WRITING_APP: mkvmerge v7.9.0 ('Birds') 64bit
      _STATISTICS_WRITING_APP-eng: mkvmerge v7.9.0 ('Birds') 64bit
      _STATISTICS_WRITING_DATE_UTC: 2015-11-04 17:32:41
      _STATISTICS_WRITING_DATE_UTC-eng: 2015-11-04 17:32:41
      _STATISTICS_TAGS: BPS DURATION NUMBER_OF_FRAMES NUMBER_OF_BYTES
      _STATISTICS_TAGS-eng: BPS DURATION NUMBER_OF_FRAMES NUMBER_OF_BYTES
      encoder         : Lavc58.91.100 pcm_s16le
[wav @ 0x558a274d1cc0] Filesize 4726645260 invalid for wav, output file will be broken
size= 4615865kB time=02:28:51.68 bitrate=4233.6kbits/s speed= 310x
video:0kB audio:4615864kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead: 0.000004%
/tmp/Test movie title.mkv

Test movie title

Erreur de segmentation

Any way to compile it for Linux?

Tried to compile it on ubuntu 19 and it doesn't seem to work.

I think it might be because the C tools on the distro are of latest version compared to what you used on your end. Feel free to correct me on that :)

Estimate the time in sample where the matched song started playing

Great little repository here 👍

I'm trying to use it to detect songs played during a radio broadcast. I'm able to find the matching result just fine but I was wondering if it would be possible to also detect where in the sample the matched song started playing. For example, in a 5 minute sample, an indexed song starts playing around the 1 minute mark, I would like to be able to detect both the matching song and the estimated time at which it started (~60 seconds).

I expanded the signature struct to also include a time property which contains the position in the indexed track when the spectral image was constructed, but I'm wondering if this would ultimately give me the results I want. Do you have any suggestions/advice on how I can extend the project to support this?

Comparing short audio files

Hi,
I'm interested in finding near-duplicate audio files. My dataset is about 3000 thousands short audio files, between 0.5 seconds to 5 seconds. Unlike Shazam, both the "target" audio (i.e. the songs in Shazam's case) and the user input are short, and both might contain noise.

Can this library help?
If so, are there any recommendations for tuning parameters?

N.B - if a file is matched to multiple other files, it's fine - I have a less efficient algorithm that can verify which match is correct. In other words, I can handle some amount of false positives, but I don't want false negatives.