Episode 1, ready to listen 22 places

Every sound you marked as cloth has been replaced with the quiet sound of the room. Nothing you called a sound effect was touched, and nothing that was still marked leave-alone when this was built was touched. Listen to it straight through, and if you want to check a particular spot there are short before and after pairs further down.

What changed since the version you heard. This one treats 22 places, the same number as before, but one of them changed shape. The one that is new is marked new further down:

14:58 to 15:04, treated as one stretch of 5.5 seconds rather than as a moment. You said it actually starts about five seconds earlier than the place that was marked, and is also a ruffle, so the whole length of it comes out together. The places that used to be listed separately at 14:59, 15:03 are inside this stretch now, which is why there is one entry here rather than several.

Everything else is built exactly the same way, from the same recording, with the same settings. And the build can now be re-made from your original recording and comes out identical, byte for byte, which it could not before.

Listen to the whole thing

The new one

16:53. This is what you would be replacing the episode with.

The one on the feed now

The same episode as it stands today, for comparison.

One thing about these two players. They play straight through, but dragging the bar to jump somewhere may not work: the server sends the whole file rather than the piece you asked for. That is why the checks below are cut as their own short files.

What was done

22 sounds, in 17 stretches. You marked 10 on 8 August and 13 in the first pass. Where two or three sit within a few seconds of each other they were fixed as one stretch, which is why there are fewer stretches than sounds. Every one of them is quieter now, between 12 and 48 dB, and not one came out louder.

Your voice was never touched. The speech is separated out first, kept exactly as it was, and put back. That was checked afterwards, sample for sample, and it is identical.

Nothing was left in by accident. The 6 sounds you called sound effects are you performing, so they were never candidates.

Nothing jumps out at you. This is the bedtime rule: no moment in the story may be much louder than the rest, because a jump after quiet narration wakes a child. The last attempt failed it at 5.6 against a limit of 5. It failed because the check was being done on the story before the volume was set, and then judged on the finished file, which are not the same thing. It is now done on the finished file, and it sits at 3.9. The limit did not move.

Check the work, place by place before and after, 17 pairs

nothing playing

What you are hearing. Both clips are the same episode at the same volume, processed the same way. The only difference between them is the replacement. Times are the times in the episode, so 3:38 here is 3:38 in your player.

Place 1 3:38 41 dB quieter

1 sound you marked here

Place 2 4:00 48 dB quieter

1 sound you marked here

Place 3 4:14 12 dB quieter

1 sound you marked here

Place 4 5:24 30 dB quieter

1 sound you marked here

Place 5 5:59 28 dB quieter

1 sound you marked here

Place 6 6:12 33 dB quieter

1 sound you marked here

Place 7 6:41 33 dB quieter

1 sound you marked here

Place 8 10:39 36 dB quieter

1 sound you marked here

Place 9 10:53 36 dB quieter

1 sound you marked here

Place 10 11:19, 11:23, 11:28, 11:32 19 to 43 dB quieter

4 sounds you marked here (they are close together, so they were fixed as one stretch)

Place 11 11:47 18 dB quieter

1 sound you marked here

Place 12 11:55 36 dB quieter

1 sound you marked here

Place 13 13:49, 13:55 32 to 39 dB quieter

2 sounds you marked here (they are close together, so they were fixed as one stretch)

Place 14 new 14:58 38 dB quieter

1 sound you marked here

Place 15 15:12 22 dB quieter

1 sound you marked here

Place 16 15:34 31 dB quieter

1 sound you marked here

Place 17 16:46, 16:51 27 to 41 dB quieter

2 sounds you marked here (they are close together, so they were fixed as one stretch)

The numbers

Length	16:53, the same episode as the one on the feed to within 36 milliseconds
Volume	-16.5 against a target of −16, which is where it needs to sit next to other podcasts in the car
No sudden jumps	3.9 against a limit of 5, measured on the finished file
Peak level	-1.8 against a ceiling of −1, so nothing distorts
Dynamic range kept	10.9. Holding the loud moments down cost 0.4 of it, which is the price of the bedtime rule
Technical score	93.0 out of 100, against 84.4 for the episode on the feed today, scored the same way
Every automatic check	passes: volume, peak, no sudden jumps, encoding, no distortion. It is also not worse than the file it would replace on any of them.

What happens next

Nothing, until you say so. This is not published and it has not replaced anything. Two separate decisions, and both are yours: first whether you accept this as the master, and separately whether it goes on the feed. Tell me which places still sound wrong and they go back into the list.