Audio manipulation within video files involves working with the audio stream at the container level, understanding how digital audio is represented, and knowing how volume adjustment works mathematically on digital audio samples. These concepts explain the different approaches to muting, removing, and adjusting audio in video files.

Within a multimedia container like MP4 or WebM, the audio stream exists as an independent track alongside the video track. Each track has its own metadata in the container header, including codec identification, timing information, and track flags. One important flag is the track enabled flag, which tells the player whether to present the track during playback. Some muting operations work by simply toggling this flag to disabled, keeping the audio data intact but instructing players to ignore it. However, this approach is unreliable since not all players respect the flag, and the audio data still occupies space in the file.

Removing an audio track entirely is a remuxing operation that copies only the video stream from the source container into a new container, omitting the audio packets completely. This produces a smaller output file since the audio data, which typically accounts for 5-15% of the total file size depending on the audio codec and bitrate, is completely eliminated. The operation is fast because no video decoding or re-encoding is required; the video stream is copied bit-for-bit. Some platforms and players handle audio-less video files differently than videos with silent audio tracks. Certain social media platforms may reject videos without an audio track or may display them differently in feeds, which is why keeping a silent audio track is sometimes preferable to complete removal.

Muting audio by replacing it with silence involves generating a stream of zero-value audio samples at the appropriate sample rate and channel configuration. In PCM (Pulse Code Modulation) digital audio, silence is represented by samples with a value of zero (or the midpoint value for unsigned formats). A silent AAC or Opus stream is extremely small because audio codecs compress silence very efficiently, representing long periods of zero values with minimal data.

Volume adjustment in digital audio is a straightforward mathematical operation applied to each audio sample. Digital audio samples are numerical values representing the amplitude of the sound wave at each point in time, typically stored as 16-bit integers (ranging from -32768 to 32767) or 32-bit floating point values (ranging from -1.0 to 1.0). Reducing volume by 50% means multiplying every sample value by 0.5, halving the amplitude. Reducing to 0% means multiplying by zero, producing silence. This linear scaling is simple to implement but does not perfectly match human loudness perception, which is logarithmic. A 50% volume reduction in linear terms sounds like roughly a 6 dB decrease, which most listeners perceive as noticeably quieter but not half as loud. True perceptual halving of loudness requires approximately a 10 dB reduction, corresponding to multiplying samples by about 0.316. Understanding this distinction between linear amplitude scaling and perceptual loudness is important when fine-tuning volume levels to achieve the desired listening experience.