Part of my work consists of adapting Audiomatarial to different frame rates. I am therefore looking for an SRC that accepts "crooked" sample rates and still delivers (very) good results. I am currently using RX. The problem is that RX often miscalculates when converting with the "official" values (e.g. 46034 kHz or 50050 kHz) and then the length is not exactly correct. I then have to do a lot of experimenting until I find the "right" sample rate.
Pro Tools does a good job here: the conversion is often (almost) perfect. (PT offers options to convert from 25 to 23.976 when importing). However, the algorithm used in Pro Tools is not the best. (We have moved our studio to Nuendo and no longer use PT anyway.)

I've tried a lot of SRCs, but many of them can't handle strange sampling rates, only the usual ones. (44.1 kHz, 48 kHz, 96 kHz, etc.)

 

Try Adobe Audition

 

I last used it in 2015.
Do you know which algorithm Audition is using?

 

you use chinese rx?

Also,

• ffmpeg -i "$input" -af aresample=$samplerate:cutoff=1 "$output" -y

 

Ask gearspace people in Mastering Thread or so.

 

I thought that iZotope RX was made by NI, which is a German company.

 

Try Saracon by Weiss

 

Thank you for your suggestion. I use SARACON occasionally. (Even though it is not one of the best in comparison.) However, SARACON is an SRC that only accepts the usual values (44.1 kHz, 48 kHz, 88.2 kHz etc.). It is not possible to enter freely selectable values such as 46.034 kHz.

 

I think you should give more details about what you are doing. Maybe you need Wavelab/audition, or NUGEN Halo Upmix, or Myriad. Maybe you could benefit from Black Magic Da Vinci, or....Sequoia?

It's hard to say with the information in the first post.

 

Thank you for your efforts.
Two years ago I tried r8brain PRO (v2.7). At that time only the standard values (44.1 kHz, 48 kHz etc.) were supported. There was no free selection of the kHz value. I therefore discarded this SRC as unsuitable for our needs.

Do I have to look for a specific version if I want to use free values?

I actually thought that the first post covered all the essentials, but it might not have been clear enough if you've never or rarely done this before.

So in detail:

We get many archive tracks, often from the DVD era. In those days, only the Dolby Digital track was saved from the DVD. That was deemed good enough. The original audio files were destroyed or lost, or nobody knows where the originals are after all this time.
Now, when a film is released on Blu-ray or UHD-BD, the DVD track is often the only one available. (Our company also produces Blu-ray and UHD BDs.) This is then transcoded to WAV (PCM) and needs to be converted from PAL (25p or 50i) to 24p or 23.976 fps. To do this we use an SRC with a "crooked" kilohertz value.

Example: 25p > 24p (difference: -4.0%) = 4800 × 24/25 = 46080 kHz.

The file will not be encoded with this value! Only the value in the metadata is changed! (The size and length of the file will not change.) Next, the 46034 kHz file will be encoded at 48 kHz. (The size and length of the file will change.)

When imported into the DAW, this WAV file is longer and sounds lower. (This is desirable for films, as it also removes the PAL pitch.) If the SRC is not accurate enough, the track will be too short or too long, and the sound will become asynchronous as the duration increases. (This means that the track is no longer lip synced.) This then has to be corrected manually, which means extra work.

 

Last edited: Sep 22, 2024

The Resample function of Sound Forge accepts any rate. You may save a preset and then use it in batch for several files in a row in Sound Forge's Batch Converter.

 

Nope, just type the value into the boxes.



I'm on version 2.1 (latest as of this writing)

 

Interesting! The Pro version does not have this field for free entry of a kHz value. In this case the free version is better than the paid one. I will try it immediately. Thank you very much.

 

Oh, wow. You're right, I just tried the Pro version. Guess I'll back up the free version, in case he changes his mind in the future

If all else fails, the version of r8brain in Reaper works for arbitrary sample rates too (and comes with dither).

 

you should be very careful about non-standard sample rates, because most playback engines and audio interfaces won't handle them

 

Playback is at standard sampling rate (usually 48kHz). The "crooked" values are only needed for the first step of the adjustment. At the latest when importing into the DAW, the file must be adapted to the sampling rate of the project. (If you haven't already used the external SRC for this.)
But you're right: many audio interfaces can only play back standard sample rates. However, in our studios we only use virtual sound cards, which usually have no problem with such exotic sampling rates.

But as I said: The normal 48 kHz sampling rate is used for playback. The "crooked" rates are only used in an intermediate step. So my method of frame rate conversion can be used with any hardware. The audio interface just has to support 48 kHz. But this should be the case for most interfaces.

 

It's not unclear. It's that your specific use may actually have a "best answer", and without the details people will just suggest whatever they use. You will end up having to review or test each method yourself anyway, and that can defeat the point of even asking. You could get that just by asking ChatGPT.

Another feature in Soundforge could be useful in this situation, if it still there. As I guessed, the idea not being Sample Rate Conversion; because any application can do that easily. Every DAW will ask when you import the files if you want the sample rate to be converted to match the main session sample rate. The "metadata" being mentioned later, indicates that what is wanted is altering sample rate *detection* values. You can separately force a value to any wave file, which will actually cause the waveform to be detected at the rate you want, with no resampling, dither, truncation, quanization errors, or silly "How do the algorithms sound different" concerns. If you want to create a clickbait style topic about how Protools algorithms sound different than another DAW; you will need to find others for that purpose.

They added the option to be able to force BPM and sample rates to files in Soundforge 4.5d or 5.0, I forget. It was added because of how the program was being changed to be used with Acid 2 or Acid 3.

 

Last edited: Sep 23, 2024

This is not (only) about the sound, but about how well the respective SRC calculates. "r8brain" does a decent job, generating (in my test) 71 milliseconds of offset between start and end. That's still not good enough, but it's better than RX. So I have marked it as "best answer". I can always change this if I find that there are even better programs. A detailed test is still pending.

 

Why does audio and video go out of sync? Wouldn't 1 second of audio = 1 second of video? Regardless of the frame rate or sample rate?
From what I have read, a 44.1 kHz audio file would fit with all the common frame rates but 24 fps... Someone correct me if I am wrong.
That is why for film and television we use 48 kHz?
Sorry for asking dumb questions, just thought it would fit here.

Unless, say the footage has a decimal value frame rate and would slowly get more and more out of sync in the playback of the project. I guess that is what the problem is here when converting digital content from one region to another, PAL to non.
If that is the case then what would be the fix? I know some cameras have a 29.97 frame rate, if I remember correctly.

I find it strange that in this day and age there seems to not be a software that allows you to accurately convert the sample rate of an audio file to a specific integer value. Especially for these types of scenarios.

 

Last edited: Sep 27, 2024