Waveforms match, audio output does not.

I am working on a video and the sample rates from two audio sources for interviews did not match, they were 48 000 and 44, 100. I upsampled the 44, 100 to match the 48 000 and swapped out the files. When I restarted Resolve, I see that visually, the waveforms now match perfectly, but the audio output does not. I restarted Resolve several times, but the same problem: wave forms match but what I hear come out of the speakers do not. It is the same interviewee, but at a different point in speech. So why on earth do the wave peaks match perfectly? Any solution? Or is this a program limitation?

How much difference, and in which direction? Does audio lag video or vice versa? Is the discrepancy a constant amount through the recording, or does it drift longer or shorter? If it drifts, does it drift by a steady amount, or variably?

If it's constant amount, and by a syllable or so, there's an option in the preferences to delay the video by x frames. Sometimes your hardware configuration can introduce slight delay in audio, video, or both. Resolve has ways to compensate.

But if you're talking about a major discrepancy, like minutes, then something else is wrong.

If it drifts, it's either tiny miscalculations adding up over time caused by the different clocks inside different devices, or it could be caused by variable framerate video encoding, it's hard to say. Often, the simple solution is to break up the video periodically, and re-sync it, and find a better recording setup/hire a sound guy the next time you shoot interviews. The whole reason timecode is a thing is this particular issue.

@Jason

Thanks for the reply. The difference is pretty significant, like wayyy off. Not a nudge, but different time range entirely. Yet, the wave forms match perfectly. I think it could be a cache issue, which I flushed, but the discrepancy is still there. I played back the upsampled file and it seems okay...so I am perplexed.

edit: I see you updated your post, which says that my footage is pretty much fugged.

Okay, I will go rebuild the project in a different NLE. No biggie, I only worked on this project for about a year in Resolve. What's a year? Live and learn, don't use Resolve for big projects.

Do you have your footage proxied by any chance? Resolve bugs out with proxies and the audio plays at different bits sometimes depending on the HZ of your audio

@dave4443 No, the footage is not proxied, it is HD footage with two mics. Most of the footage syncs up, but the couple of interviews I had to upsample do not. If I take just those interviews and rebuild in another NLE, I suspect they will sync because it is two of the same samplerates.

Update: I swapped the files again and noticed that there are still issues. This time I re-rendered the file to be more accurate, in Reaper, this time. There are some effects in the region that remain, like 'amplify', but the wave forms do not show this. Let me clarify, when I first put the wav file, the signal was low, so I amplified it. This time I swapped with a file that was amplified first before inserting. The wave file looks fine, but the audio is obviously clipped. Normally you can see clipping in the wave form. I played the file on its own, and it sounds absolutely fine. However, on the Resolve time line it sounds clipped and distorted.

I've never had a problem using 44.1 audio files in a 48 project.

Have you tried simply using the original audio?

I played the file on its own, and it sounds absolutely fine. However, on the Resolve time line it sounds clipped and distorted.

Do the meters peak in Resolve? The file might be fine, but it’s still totally possible to overdrive. The more info you can provide, screenshots, captures, etc, the more people here will be able to help. There are a few Fairlight experts around here.


Sent from my iPad using Tapatalk

44.1khz audio files should be no problem as Resolve sample rate converts them to 48khz automatically. As Jason says, some screenshots would be helpful.

The 44 100 and 48 000 wave files will not match, they are off. There is also a discernible echo in playback, and it makes sense that it is so. Swapping the files out to matching sample rates brings pixel perfect waveform match, but the audio is different, like two voices speaking about different topics. There is no screenshot that can let you hear this

The 44 100 and 48 000 wave files will not match

You're using two files of the same thing?

Yes, the exact same interview, but from two sources.

Soooo...a different mic fed each?

How's your mom?

Doing well, thanks. Saw her a couple nights ago.

Different mics? What hardware was used to record each?

Your mom is super cool.

I like her.

Different mics? What hardware was used to record each?

The situation is resolved, no pun intended. My error was that instead of using 'replace selected clip' in the media folder when swapping files, I just deleted the files and replaced the up-sampled files with the same name in the folder. While visually the wave forms were the same, the playback was from the prior 44,100 file. When I clicked the wav file, the readout was 44, 100, though only 48,000 files were in the folder. I then deleted the 48, 000 files and saw a red rectangle on the timeline that indicates files were missing. I then replaced the original 44, 100 files AND THEN did 'replace selected files'. Once I did so, the swap occurred without a hitch. Only one interview needs a 'nudge' now.

For those who insist that Resolve can mix 48 000 and 44, 100 samplerates, that may be so, but in my case, it does not as you can see the two wave forms in the image below. The top is 48,000 and the bottom is 44, 100, which is why upsampling and swapping is necessary. This is Resolve 17.2.2 build 4.

Scott, I've sent you a PM. Upsampling is not needed as the sample rate, whether 44.1k or 48k, does not change the length of a file and Resolve does automatically sample convert when playing the files. I think there is another problem here.
For others that are interested here is a pic of the same audio file in Resolve at two different sample rates. They match, which is what I would expect them to do.

The up-sampling and changing the source files helped, but I still have a couple of interviews that are not fully matched, despite being the same sample rates. When I move them on the timeline, they will not match. These need a 'nudge'. How would I do that?

You could try using the Speed Change in the Inspector to make the track the same length as the original without altering the pitch. Failing that it's an editing job cutting up the track and moving sections to bring them into sync.

Out of interest, what was the problem track originally recorded on?