Recherche avancée

Sur d’autres sites (12802)

• Pydub FFMPEG issue [closed]

  14 janvier, par Nikolai van den Hoven

  I am attempting to use FFMPEG with Pydub to create a program that chops .mp3 files into different words, each contained in their own .mp3 file, but when I run the script I am getting the following error :

  


  PS C:\Users\nik> & C:/Users/nik/AppData/Local/Microsoft/WindowsApps/python3.12.exe "d:/Python/Word Splitter.py"
C:\Users\nik\AppData\Local\Packages\PythonSoftwareFoundation.Python.3.12_qbz5n2kfra8p0\LocalCache\local-packages\Python312\site-packages\pydub\utils.py:170: RuntimeWarning: Couldn't find ffmpeg or avconv - defaulting to ffmpeg, but may not work
  warn("Couldn't find ffmpeg or avconv - defaulting to ffmpeg, but may not work", RuntimeWarning)


  


  This is the code I am using.

  


  import os
from pydub import AudioSegment
from pydub.silence import split_on_silence
import speech_recognition as sr
AudioSegment.ffmpeg = r"D:\Python\ffmpeg\bin\ffmpeg.exe"
def mp3_to_words(mp3_file, output_folder):
    # Ensure output folder exists
    os.makedirs(output_folder, exist_ok=True)

    # Load MP3 file
    print("Loading audio file...")
    audio = AudioSegment.from_mp3(mp3_file)

    # Split audio into chunks using silence detection
    print("Splitting audio into chunks...")
    chunks = split_on_silence(
        audio,
        min_silence_len=200,  # Minimum silence duration in ms to consider as a split point
        silence_thresh=audio.dBFS - 14,  # Silence threshold relative to average loudness
        keep_silence=100  # Retain some silence in chunks
    )

    recognizer = sr.Recognizer()

    for i, chunk in enumerate(chunks):
        print(f"Processing chunk {i + 1}/{len(chunks)}...")

        # Save the chunk temporarily
        temp_file = os.path.join(output_folder, f"chunk_{i}.wav")
        chunk.export(temp_file, format="wav")

        # Recognize words in the chunk
        with sr.AudioFile(temp_file) as source:
            audio_data = recognizer.record(source)
            try:
                text = recognizer.recognize_google(audio_data)
                words = text.split()

                # Export each word as its own MP3
                word_start = 0
                for j, word in enumerate(words):
                    word_duration = len(chunk) // len(words)  # Approximate duration per word
                    word_audio = chunk[word_start:word_start + word_duration]
                    word_file = os.path.join(output_folder, f"word_{i}_{j}.mp3")
                    word_audio.export(word_file, format="mp3")
                    word_start += word_duration

            except sr.UnknownValueError:
                print(f"Could not understand chunk {i + 1}.")
            except sr.RequestError as e:
                print(f"Could not request results; {e}")

        # Clean up temporary file
        os.remove(temp_file)

    print(f"Processed {len(chunks)} chunks. Word MP3s saved in {output_folder}.")

if __name__ == "__main__":
    input_file = input("Enter the path to the MP3 file: ").strip()
    output_dir = input("Enter the output folder path: ").strip()

    mp3_to_words(input_file, output_dir)


  


  I have added the Base FFMPEG folder and the bin folder within it to Windows PATH
My PATH variable on Windows 11,
But it does not show up in the variable when I typed PATH into cmd

  


• ffmpeg is making my audio and video frozen and I don't know why

  17 avril 2024, par Sdpro

  I'm using bunjs runtime to execute ffmpeg as terminal code but I don't know if my code is typescript code is wrong or ffmpeg is wrong
and I'm using json file to get the clips correctly

  


      let videos = 0;
    let stepsTrim = "";
    let concatInputs = "";

    for (let i = 0; i < 40; i++) {
        if (unwantedWords[i].keepORdelete === true) {
            stepsTrim += `[0:v]trim=0:${
                unwantedWords[i].start
            },setpts=PTS[v${i}];[0:a]atrim=0:${
                unwantedWords[i].start
            },asetpts=PTS-STARTPTS[a${i}];[0:v]trim=${unwantedWords[i].start}:${
                unwantedWords[i].end
            },setpts=PTS[v${unwantedWords.length + i + 1}];[0:a]atrim=${
                unwantedWords[i].start
            }:${unwantedWords[i].end},asetpts=PTS-STARTPTS[a${
                unwantedWords.length + i + 1
            }];`;

            concatInputs += `[v${i}][a${i}][v${unwantedWords.length + i + 1}][a${
                unwantedWords.length + i + 1
            }]`;
            videos += 2; 
        }
    }

    stepsTrim = stepsTrim.slice(0, -1);

    await $`ffmpeg -hide_banner -i ${videoRequirements.output} -filter_complex "${stepsTrim},${concatInputs} concat=n=${videos}:v=1:a=1[outv][outa]" -map "[outv]" -map "[outa]" -c:v libopenh264 -preset slow -c:a mp3 -vsync 1 -y ${removedUnwantedWords}/fastAf.mp4`;


  


  at the end after everything was done :

  


  warning
[vost#0:0/libopenh264 @ 0x558f80ea1dc0] More than 1000 frames duplicated.9kbits/s dup=110 drop=1 speed=0.831x    
[out_0_0 @ 0x558f8100a880] 100 buffers queued in out_0_0, something may be wrong. dup=1064 drop=1 speed=1.43x    
[out_0_1 @ 0x558f8100af80] 100 buffers queued in out_0_1, something may be wrong.
[out_0_1 @ 0x558f8100af80] 1000 buffers queued in out_0_1, something may be wrong.
I can't figure out why ffmpeg is sometimes making the audio + video work and sometimes not
[enter image description here](https://i.stack.imgur.com/PicaA.png)


  


  [
  {
    "word": "Hello",
    "id": 0,
    "keepORdelete": false,
    "start": 0,
    "end": 9.06
  },
  {
    "word": "guys,",
    "id": 1,
    "keepORdelete": false,
    "start": 9.06,
    "end": 10.2
  },
  {
    "word": "there",
    "id": 2,
    "keepORdelete": false,
    "start": 11.76,
    "end": 12.06
  },
...


  


  I have tried commands from many types of ffmpeg commands changing the code and I can't seem to get the audio and video right

  


• I am trying to develop On demand and live streaming video analysis using java.?

  2 avril 2020, par PATEL UJASH

  After a lot of R&D I got something for video analysis.I am gonna use Kafka, spark and openCV. But i am suck in how to deliver streaming on demand and live streaming video. I got FFMPEG help to capture video from different devices or you can stream any Video File too. FFMPEG also help in Video encoding and decoding also. But after encode the video we need to send that data Into any streaming engine like wowza streaming engine, Red5 or AWS media elemental services. I watch some video they have used OBS software for stream.

  



  What if i want to stream it using FFMPEG and publish the video streaming on RTMP link. That link are generated by wowza streaming engine or aws media elemental service like rtsp://[wowza-ip-address]:1935/[application]/[stream-name] ?

  



  How to deliver different video frames resolution according to bandwidth ?.