Sound Forge 10 (PC) file save tasks that cause transcoding or don't...

murrayatuptown wrote on 4/22/2018, 9:50 PM

Hi:

I bought Sound Forge Audio Studio v.10 at the suggestion of a friend for simple editing of audio from .mov camera videos.

It has sufficient video capability for me, actually more than I anticipated, but I am plagued by not understanding which actions I apply result in nearly identical file size (minimal increase) and other times massive file increases. I saw a similar discussion under another product's category.

My question is at the end of this, but I have explored many different software paths and like everyone else sometimes have different needs, so I explained my twisted path...

Basically trying to deal with horrible audio quality in every form of digital camera (still or video) I have recorded musical performances with, I have attempted editing audio with it remaining attached, separately, then re-muxed (other software before discovering SFAS), and attaching (dubbing?) separately recorded files from a digital audio recorder.

Detaching and re-attaching audio was causing significant file length increases in other programs...far worse than 'clock drift'...like almost doubling the timeline. Three different cameras.

SFAS is the first program that actually showed me what was happening to my files...they were imported at the nominal 29.97 (30) fps I expected, but when saving the results, the frame rate was some totally unfamiliar number like 18.64 FPS (I don't remember exactly). SFAS would display the 'different' FPS number, but also allow me to type the correct numeric value back in and save it successfully. Several other 'freeware' programs never even recognized the frame rate was being changed. It's definitely NOT a SFAS problem. SFAS just reveals the frame rate corruption.

The guy who recommended it is a software engineer in a totally unrelated industry but he said he used this program instead of a free program because 'you get what you pay for' and it's not his area of interest to dig into audio/video containers. What he did say, however, was that, particularly with consumer grade or cameras that have video capability as an 'extra' feature, there are too many hardware and software protocols for manufacturers to comply with 100% in all situations...they do what they need to for their requirements, for economic reasons. He said basically if you 'break' a file's interleaving with the kinds of experiments I was trying, all bets are off.

I could never find anyone who experienced the same audio frame rate corruption problem because people who have professional needs buy the proper industry standard hardware and software they need instead of hacking & bandaging things as I was.

A separate detour was surgically removing the electret mic from two cameras and hacking an external mic jack onto the cameras...to improve what audio was captured to minimize the post-processing necessary. That worked but introduced new issues.

So, what I would like to learn is why I am sometimes able to save (or was it export?) as .mov, files I import as .mov and perform audio editing, or splitting or trimming operations on, without massive file size (like 75-100x) increases. It's like some operations cause transcoding and others don't. It seems like all is well until I save as mov. I assume I MUST save-as after reattaching the video or the result won't get saved.

I typically save the audio as .wav after detaching the video and performing audio edits. I might have tried saving audio-only files as .mov because it was in the save-as menu, but think it produced a massive file so I assumed it was wrong.

 

Thank you

Murray

Comments

rraud wrote on 4/24/2018, 9:12 AM

DSLR's are notorious for awful sounding audio, even with a top shelf mic and preamp, it's still gonna suck. Most folks use a separate portable recorder (a Zoom or Tascam for instance) and replace the camera's audio using PluralEyes or manually syncing the waveforms (a clap slate helps). Vegas is better for this though. For long form work, drift between the devices is normal, so realignment every 10 minutes or so is necessary.

murrayatuptown wrote on 4/24/2018, 11:19 AM

Thanks...

 

yeah...poor is an upgrade from horrible!

I forgot the mention I did a Linkwitz mod twice on the 4mm ECM, under a microscope...2nd time to change my incorrectly estimated source resistor.

the 3rd one was an upgrade (where’s the laughing emoticon?) to a 6 mm one, hanging from wires after carving the original one out from the front...I figured my luck would run out if I kept taking the camera apart!

The audio was as good as I could expect from a probably 8 kHz sampling rate on a point&shoot camera codec chip but all the ECM’s were omnidirectional. Apparently the chipset had companding or at least compression and it would pick up everyone in the room talking or clapping. A fun exercise but another set of frustrations.

I never thought at the start I’d go as far as I did...added a blocking capacitor and XLR jack to protect the camera from phantom power...try to make it idiot-proof...

Next up was a hypercardioid snare mic that seemed like a dumb purchase earlier as it had wrong hardware for a mic stand...but easily adaptable, pretty deaf (great for overload insurance) and good pattern so the loudmouth sitting next to me inaudible.

No A/V synch problems but still had the mediocre codec...still better than unusual. Distortion and wind-noise significantly improved...

I still have simple tasks, hiwever, like splitting files into multiple files and saving-as, and audio enhancement then save as.

Lately I’ve been using the iPhone whuch is pretty good except in really loud situations.

So I still need to learn what can be done to audio or video without causing the transcoding...save-as may be guilty of this...I’m not certain...could be operator error or poor choices/methods.

 

thanks

murrayatuptown wrote on 4/24/2018, 11:20 AM

Unusual should say ‘unusable’

murrayatuptown wrote on 1/29/2019, 9:18 PM

I am still having massive file size increases, so it must be operator error.

New phone, creates .mp4 videos. Processing audio in the videos with Sound Forge 10, then saving as .mp4 seems to save only the audio portion of an mp4 video, but if I save the mp4 as .mov, it saves a video, but many times larger (like 86x larger: 440 MB became 38 GB!

'Custom' tab allows reducing size and quality to reduce size, but results are still at least twice the size of the original source video

Is there some parameter I am not setting correctly to simply process audio in video? Do I need to detach video, edit audio, then reattach video?

 

Thank you

rraud wrote on 1/30/2019, 3:09 PM

That is because SF is re-rendering the video likely at a higher bitrate and other factors affecting file size. Re-rendering video should be avoided unless the picture is changed or needs to be in a different format The Magix versions of SF AS (12 &13), allow replacing the audio w/o re-rendering the video. This is typically known a 'muxing'. There are free third-party muxing utilities available though. I use MeGUI,TS Muxer and AvideMux, However, I am not sure if any of the above support MOV files. But you can probably just change the file's extension to .mp4 if it's a MOV.

murrayatuptown wrote on 1/30/2019, 5:37 PM

Thank you.

I had used a couple things years ago...not sure I remember, maybe MP4BOX & something similar, when I was trying to dub externally recorded audio. I was having huge file duration (time, not data) increase, worse than clock drift. SF was the first program I ever encountered that showed me the frame rate changed during audio edits. That s/w guy I mentioned earlier told me it's not the fault of the s/w but my 'breaking' the video container in some way the original camera's management of audio & video interleaving (management of all the framing, sync, etc) didn't anticipate...they just needed to make their camera do what THEY wanted.

SF allowed me to overwrite the misinterpreted video frame rate (imported at 29.9997 fps but was something drastically different when saving-as...yes, the re-rendering you mention (or re-encoding, transcoding, whatever) became the villian.

While external recorder device (I have a Tascam PCM one somewhere) is really a better way, I found editing the video file's audio without separating and re-muxing (attaching) the audio & video was my brute-force way of preventing the time duration changes)...but I always get blind-sided by the unexpected re-render/encode.

I started looking at/test-driving other Magix products today to see if something else newer would help. I am also checking out Video audio cleaner lab (forgot the name) product because I don't know if SF is really what I need...it was an accidental discovery but quite helpful. v. 12 or 13 might be the simplest change without relearning everything (but using it once or twice a year means...re-learn anyway). I think all I really have a need for is the audio edits minus re-render/encode...not in need of video titling, non-linear editing, etc.

The Pixel phone I just used records at 44100 bps for audio, so that's a huge improvement from the old cameras. I also learned I surely do not need 1080 HD or 4k UHD for web publishing! Impractically huge files! So, I'm working on a 'dumbing them down' workflow. I might go back to trying external microphones or a mixer to feed the phone audio input, but in many cases the smart phone audio has been pretty impressive (despite non-ideal microphone placement-sheer luck basically).

To be honest (and wordy), other than your pointing out SF 12, 13 are able to edit audio without re-rendering the whole video, that's a hard thing to discover about a s/w product from the sales literature!

Thanks for your patient reading and ongoing responses...one thing I learned asking 'experts' is that they buy what they need/what works instead of pursuing alchemy with random open source s/w etc. Avoiding the rabbit hole, as it were.

I do waste a lot of time figuring out many wrong ways to do this kind of work(hobby, no doubt), but I prefer that to bowling, or whatever...

Thank you.

Murray

murrayatuptown wrote on 1/30/2019, 5:51 PM

Only reason I switch between .mov and .mp4 is if whatever s/w I'm negotiating with seems to force me when I want to complete an uncooperative function.

Not sure I am correctly repeating this, but I was told (sometime ago) .mp4 and .mov extensions refer to 'video containers' rather than the codec 'function' (like h.264 and so on). An .mov and .mp4 file could hypothetically have the same encoding but are put together differently. That was why two different cameras that both created .mov files or .mp4 files were still too different in certain circumstances (like trying to put one camera's .mov file onto an SD card with the other camera's .mov files). I think we're not supposed to be 'looking behind the curtain'...just using what people created for us! (No user-serviceable data inside warning).

I was not meant to be a software guy but apparently am a 'Butterfly Effect trespasser'...I get inside, wreak havoc on myself, and...spend tens of hours instead of tens of $...I just don't seem to learn from my mistakes!

 

rraud wrote on 1/30/2019, 6:05 PM

In audio, a different sample rate than what is it was recorded as causes faster or slower playback.Drift is always a problem, but usually only noticeable when the takes are long, but then sometimes you just get lucky. For perfect long term sync a master word clock/genlock is needed so all the slaved devices run at the exact same speed, timecode or not.