Recherche avancée

Sur d’autres sites (6091)

• avutil/x86/pixelutils : Empty MMX state in ff_pixelutils_sad_8x8_mmxext

  1er novembre 2023, par Andreas Rheinhardt

  avutil/x86/pixelutils : Empty MMX state in ff_pixelutils_sad_8x8_mmxext
  We currently mostly do not empty the MMX state in our MMX
  
  DSP functions ; instead we only do so before code that might
  
  be using x87 code. This is a violation of the System V i386 ABI
  
  (and maybe of other ABIs, too) :
  
  "The CPU shall be in x87 mode upon entry to a function. Therefore,
  
  every function that uses the MMX registers is required to issue an
  
  emms or femms instruction after using MMX registers, before returning
  
  or calling another function." (See 2.2.1 in [1])
  
  This patch does not intend to change all these functions to abide
  
  by the ABI ; it only does so for ff_pixelutils_sad_8x8_mmxext, as this
  
  function can by called by external users, because it is exported
  
  via the pixelutils API. Without this, the following fragment will
  
  assert (on x86/x64) :
  
      uint8_t src1[8 * 8], src2[8 * 8] ;
  
      av_pixelutils_sad_fn fn = av_pixelutils_get_sad_fn(3, 3, 0, NULL) ;
  
      fn(src1, 8, src2, 8) ;
  
      av_assert0_fpu() ;
  [1] : https://raw.githubusercontent.com/wiki/hjl-tools/x86-psABI/intel386-psABI-1.1.pdf
  Signed-off-by : Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
  

  • [DH] libavutil/x86/pixelutils.asm

• Ffmpeg - transform mulaw 8000khz audio buffer data into valid bytes format

  24 décembre 2023, par Bob Lozano

  I'm trying to read a bytes variable using ffmpeg, but the audio stream I listen to, sends me buffer data in mulaw encoded buffer like this :

  


  https://github.com/boblp/mulaw_buffer_data/blob/main/buffer_data

  


  I'm having trouble running the ffmpeg_read function from the transformers library found here :

  


  def ffmpeg_read(bpayload: bytes, sampling_rate: int) -> np.array:
"""
Helper function to read an audio file through ffmpeg.
"""
ar = f"{sampling_rate}"
ac = "1"
format_for_conversion = "f32le"
ffmpeg_command = [
    "ffmpeg",
    "-i",
    "pipe:0",
    "-ac",
    ac,
    "-ar",
    ar,
    "-f",
    format_for_conversion,
    "-hide_banner",
    "-loglevel",
    "quiet",
    "pipe:1",
]

try:
    with subprocess.Popen(ffmpeg_command, stdin=subprocess.PIPE, stdout=subprocess.PIPE) as ffmpeg_process:
        output_stream = ffmpeg_process.communicate(bpayload)
except FileNotFoundError as error:
    raise ValueError("ffmpeg was not found but is required to load audio files from filename") from error
out_bytes = output_stream[0]
audio = np.frombuffer(out_bytes, np.float32)
if audio.shape[0] == 0:
    raise ValueError(
        "Soundfile is either not in the correct format or is malformed. Ensure that the soundfile has "
        "a valid audio file extension (e.g. wav, flac or mp3) and is not corrupted. If reading from a remote "
        "URL, ensure that the URL is the full address to **download** the audio file."
    )
return audio


  


  But everytime I get :

  


  raise ValueError(
    "Soundfile is either not in the correct format or is malformed. Ensure that the soundfile has "
    "a valid audio file extension (e.g. wav, flac or mp3) and is not corrupted. If reading from a remote "
    "URL, ensure that the URL is the full address to **download** the audio file."
)


  


  If I grab any wav file I can do something like this :

  


  import wave

with open('./emma.wav', 'rb') as fd:
    contents = fd.read()
    print(contents)


  


  And running it through the function does work !

  


  So my question would be :

  


  How can I transform my mulaw encoded buffer data into a valid bytes format that works with ffmpeg_read() ?

  


  EDIT : I've found a way using pywav (https://pypi.org/project/pywav/)

  


  # 1 stands for mono channel, 8000 sample rate, 8 bit, 7 stands 
for MULAW encoding
wave_write = pywav.WavWrite("filename.wav", 1, 8000, 8, 7)
wave_write.write(mu_encoded_data)

wave_write.close()


  


  This is the result : https://github.com/boblp/mulaw_buffer_data/blob/main/filename.wav

  


  the background noise is acceptable.

  


  However, I want to use a FFMPEG instead to avoid creating a tmp file.

  


• avcodec/hashtable : Only free buffer if there is buffer to free

  3 juin, par Andreas Rheinhardt

  avcodec/hashtable : Only free buffer if there is buffer to free
  Reviewed-by : Emma Worley <emma@emma.gg>
  
  Signed-off-by : Andreas Rheinhardt <andreas.rheinhardt@outlook.com>
  

  • [DH] libavcodec/hashtable.c