Splitting audiofile / inserting new audiofile into the middle of existing

Transloadit
1

Hi there,
I’m very pleased with audio concat, using it to append and prepend audio files. However, I’m faced with the challenge of inserting a file in the middle of another. At a given time stamp. So I guess the flow could be to split audiofile A into pieces: A1, A2 - append the new file B onto A1, append A2 onto (A1B) resulting in (A1BA2). Anyone know if this can be achieved, or if there’s an entirely different way? Thanks so much in advance :slight_smile:

2

Hello, we can achieve your use case using two /audio/encode Robots with some custom ffmpeg code to cut the audio file into two halves; then, using our /audio/concat Robot, you can sandwich a third imported audio file between your previously halved audio file.

The one shortfall of this method is that you need to specify where your audio file’s middle point is (see -ss and -t flag).

I used a minute-long audio file to test this template.

{
  "steps": {
    ":original": {
      "robot": "/upload/handle"
    },
    "audio_imported": {
      "robot": "/http/import",
      "result": true,
      "url": "https://demos.transloadit.com/inputs/powered-by.mp3"
    },
    "first_half": {
      "use": ":original",
      "robot": "/audio/encode",
      "result": true,
      "ffmpeg": {
        "ss": "00:00:00",
        "t": 30
      },
      "ffmpeg_stack": "v3.3.3",
      "preset": "mp3"
    },
    "second_half": {
      "use": ":original",
      "robot": "/audio/encode",
      "result": true,
      "ffmpeg": {
        "ss": "00:00:30",
        "t": 30
      },
      "ffmpeg_stack": "v3.3.3",
      "preset": "mp3"
    },
      "merged": {
        "robot": "/audio/concat",
        "preset": "mp3",
        "ffmpeg_stack": "v3.3.3",
        "use": {
          "steps": [
            {
              "name": "first_half",
              "as": "audio_1"
            },
            {
              "name": "audio_imported",
              "as": "audio_2"
            },
            {
              "name": "second_half",
              "as": "audio_3"
            }
          ]
        }
      }
    }
}
2 Likes
3

Thanks so much Warden! This is brilliant. Will get to it first thing tomorrow! Now that I have you :slight_smile: I don’t suppose it would be possible to have ffmpeg detect the insertion spot by “listening” for at cue, like a high pitched beep, out of our ears range? Would make this work flow significantly more automatic :slight_smile:

4

Unfortunately, as far as I’m aware this isn’t an available feature FFmpeg has. Funnily enough, I looked for an API to do this a few weeks ago and didn’t find anything.

5

Thanks again for your help. I look into it and post here if I discover something useful.

6

I’ve successfully integrated this on my server and all is looking good! One question though: I can’t seem to get milliseconds to work, can you confirm this is not supported? A second is a pretty long span under certain circumstances.

7

So to represent millisends you need to use a decimal value. I just successfully tested with the flag and value "ss": "00:00:30.50".

8

I don’t know how I missed that, thanks :slight_smile: