New Bach Harpsichord Album

Italian single-manual instrument from Orchestral Tools Berlin Harpsichords collection. Only plugin used is some PhoenixVerb (main pair: large hall). Loudness to c. -16 LUFS momentary max. Recorded direct to audio in Ardour (Win10). Recorded “live” with each track in one take and no edits…


Sounds great! Good luck with the rest of the project!

I was trying to do some classical work with Ardour recently, but have given up after having worked on one track only. I fixed initial crashes, and was able to finish the track, but it was actually the interface that was slowing me down so much, that I went back to the mainstream solution.

Interestingly, now, as I am about to master the project, to my ears the track I did with Ardour seems to stand out. Is there any reason that might be true, or can it be coincidence? I don’t find logical explanation.

What was it about Ardour’s interface that was slowing you down so much?

Are you saying that the screen layout etc was too different to what you were used to? If so, stick with it as it only takes one or two projects to get up to speed. For me, it is now as fast as other DAWs with some really nice features that I don’t easily find elsewhere. For example, multiple simultaneous export formats with LUFS normalization, dedicated CD track markers that automatically “stick” to the proper frames and, transparent “objects” that make classical editing so easy that I don’t need a dedicated crossfade window (although some basic source-destination editing function would be nice!). With version 6 coming down the pipeline with all the good latency compensation stuff, it is going to give some serious competition.

Ardour should sound like every other DAW that aims to be neutral. However, I find that the SRC on export is fabulous (Secret Rabbit Code?) and there’s something deeply satisfying to complete a project with open source software. While some of it is in my head, I probably do think better about gain-staging etc after watching all the cool Mixbus tutorials so that might translate to why I often enjoy the final exports from Ardour. Well, that and the cool export graphs that make me feel part musician, part scientist :wink:


The main problem with editing classical music - the thing I do mostly with Ardour - is the length of the files. A live concert alone may take up to two hours and a rehearsal (from which I take good parts as edits) lasts typically three hours. When I import all of that to Ardour I get a timeline of some 5 hours or so. Now, the edits most of the time consist of several eighth notes - less than a second - and source and destination can be hours apart on the timeline. I solve this problem with a lot of precisely named markers, but I will admit that zooming in and out all the time takes a lot of time and concentration (there are few dozen edits per one concert). I know that right now is not the best time to even mention additional features (and I apologize for that), but the simplest and most basic solution I can think of would be to enable selecting a portion of a file for import in the “Add Existing Media” dialog by adding two buttons, “start selection” and “end selection”, preferably bound to some keys. That way the focus would remain on the “destination” and the edits would be imported sparingly from another window (or dialog).

It is certainly the case that Ardour does not support the “source/destination” style of editing that many “traditional” classical music editors seem to favor. It’s on our list to add, but not a particularly high priority (this style of editing is fairly unusual in the overall scheme of things, though important nevertheless).

When I’ve seen live sessions with classical performers, I am constantly amazed by the general assumption that they will fail to play the piece correctly, and the desire to just grab a few bars at a time and then edit it later. Quite remarkable if you’re more used to recording jazz or other improvised music performers!

So, actually, my favorite experience with source/destination has been in Sequoia in a single-window mode (there is also a split view). I have done the same as you but with three recordings of the same concert and happily shuttle back and forth along the timeline. I color each concert object so it is clear where edits are sourced from. I don’t find it a problem. Pyramix forces you (I think) into a split view and it just seems like a bit of a hassle although with that comes easier shortcuts given the independence of each view. I made some Reaper scripts a while back building upon another user’s work and there also scrapped the two project tabs idea given that it feels even more clunky than split view. Perhaps I’m a unique case! What I liked about Sequoia is that it required zero setup to do source-destination edits. And the crossfade editor is magnificent though I can understand how problematic it would be in Ardour given the multiple layering idea. I passed on a detailed ideas to Ben Loftis several times but haven’t heard back about anything moving forward.

The more I record my own music, I try to do less and less editing and just enjoy it as a snapshot of that moment. I try to encourage other groups to do the same especially for live concerts. I can make absolutely invisibly edits in Sequoia without breaking a sweat after years of practice but something is always changed in the emotional content for whatever reason.

Sequoia (and Samplitude to which it is closely related) are definitely among the most highly evolved source/destination editors out there.

Samplitude doesn’t have the crossfade editor or source/destination stuff but the two products are getting more similar with every release. DDP import/export was added in a recent version and therefore it makes it more and more difficult to justify the high upgrade price for two classical editing features that in of themselves have stayed the same every Sequoia release! This is why I said to Ben Loftis that there is a chance for Ardour to become something of a haven for classical engineers. My Reaper scripts were quite popular but there’s nothing like a baked-in source/destination/4-point editing function. It is true that source/destination does really need a crossfade window to go alongside it. Most classical editors, AFAIK work by making 4-point edits by ear (often good enough) but if there’s a need to tweak, the crossfade editor is priceless with its two-lane view so you can precisely line up waveforms then move/re-size/shape the crossfade as you see fit (while affecting all tracks if you so wish and ripple editing in both directions so no other edits get messed up).

This all being said, it really is easy in Ardour to line up good edits especially given the transparency when dragging and the default short fades which are probably very close to the default Sequoia S-D fade type. It comes down to time-saving, honestly. It can often take only four keyboard presses while listening back and then one to make the actual edit happen compared with goodness knows how many mouse clicks, drags etc in other DAWs. Try replacing two measures of a classical recording and it quickly becomes apparent the time-savings…Of course, I’m preaching to the converted :wink:

I have heard some fantastic things about Sequoia, too. But the main reason I use Ardour and have been using it for more than a decade is that it’s currently the most capable and stable DAW to run on Linux, the OS of my choice. BTW, Sequoia costs 3000€… and runs on Windows only, which introduces all kinds of maintenance troubles I am not willing to put up with after so many years of using Linux exclusively. One of the most beautiful things about open source operating systems (Ardour is in FreeBSD repositories, too - just saying… not meant to encourage anyone…) is that you can install minimal system and then add applications needed for one purpose only, in our case audio production, and have dedicated machine that will not change, slow down or deteriorate in performance for years; a sort of appliance.

But really, I don’t have that much trouble editing stuff in Ardour as to consider alternatives. There are some - perhaps not so easy to find - features that make editing fast and efficient. The only one real problem for me is not having possibility to view or browse all of the resources in one screen (be it in one window, multiple windows or tabbed window) without having to search for them all the time.
Some of the features that make my workflow efficient are these:

  1. Markers - you can add them while listening by selecting playhead as edit point and pressing “Tab”. You can also rename them to make searching really easy.
  2. Track grouping - A must for multichannel recordings. Enables you to edit, cut , copy, move multiple vertically stacked tracks simultaneously.
  3. Selection keybindings, “,” to start selection and “.” to end selection (playhead as edit point) also make it possible to select only by listening and cut (Ctrl+x), copy (Ctrl+c) and paste (Ctrl-v) with very few if any mouse clicks (use “P” to place playhead to desired position or “`” - the key above “tab” - to change edit point to mouse). There is also “S” key for splitting track which I also use very often .
  4. Maybe the most important keybinding for editing long and very much spliced track is the one that selects all regions to the right, “Ctrl+Shift+e”, which I rebound to “Ctrl+<” because I use it so often and it is very convenient on my keyboard layout. This binding prevents you to mess up previous edits.

I hope some of this help and I wish you happy editing and success with your fine project!

This is great! For point 4, could you not use ripple edit mode for the same effect? For point 3, after you have copied and pasted the “source” into the “destination” how do you go about tweaking the fade and position of audio? What do you do, as is often the case in classical, when the replacement audio is a slightly different length? This is where real 4-point editing works wonders…

Please don’t misunderstand me. I love Ardour and Mixbus and only continued to use Sequoia (and now moved to Pyramix given the price differential) when I occasionally record album sessions when I know there will be many edits to accomplish (with Pyramix having the benefit of the rather fine Hepta Apodizing SRC and Album Publishing). I use Ardour in Windows for a lot of my VSTi work given that the two harpsichord collections I use on a regular basis run in Kontakt. For everything else, I fire up my AntiX Linux partition and do everything from start to finish with open source software and plugins, from Ardour through to Andreas Ruge’s DDP Tools, Sound Converter and Kid3 for any extra tagging. The x42 plugins see regular use, especially the LUFS metering one. The one plugin that breaks that rule is my limiter on Linux - LoudMax - but happy to receive suggestions for open source transparent true-peak limiters of similar quality.

Yes, that is sad reality.
My primary occupation in life is being a bassoonist in my nation’s finest philharmonic orchestra (sometimes I record my colleagues, as a side-job) and while I did play concerts that were both energizing AND neatly performed (or, as we say “just press rec.” concerts), there is no guarrantee that the next one will be such. Large orchestra concerts are plagued with many problems and challenges. We are playing on the instruments which basic design didn’t change for hundreds of years and are negatively affected by stupid things such as changes in air humidity and temperature. The repertoire we are playing is often pushing to the limits of what is possible in the best of circumstances - for example playing extremely quiet in extreme registers of reed woodwinds or playing some virtuoso part on horn or trombone, or fast passages with unusual scales while breaking octaves on any instrument. There can be up to hundred people playing without clear idea who is playing lead and who is accompaniment at any given moment (and that changes from one bar to another) while they sit so far apart that they don’t hear each other, but rely on one person’s gestures which are most of the time very vague… then there is this concentration issue when all goes well, suddenly you experience total blackout for a fraction of a second (usually when trying to focus very hard for more than half an hour) which is enough to produce “audible artifact”… My friends who are both jazz or latino/samba/rumba and classical music players are way more likely to play something wrong when they play classical then otherwise. I am not saying that jazz is inferior music, but I’d say it is more in the line with human nature, creativity and inspiration (any unusual move is not interpreted as mistake, but as voicing and is considered new and interesting, and rightly so, I’d add).

And so, when recording session come, producers always want at least two extra takes in case something is not as written in the score, even if they didn’t notice it on “live listening”.

But in my experience, it is not nearly as bad, last time when we recorded an album for a major German label, we did it with a general rehearsal, two concerts and one correction session, three days altogether.

I hope this demistifies the matter a little bit.

I don’t use ripple edit because, at least for me, it introduces more problems than it solves. I just delete a selection I want out and copy edit from another place and trim regions and crossfades manually. I want it to sound musically so I don’t leave edits in hands of machine algorithms, but trim it by ear :slight_smile: I In the last couple of years everyone is recording video + audio so I need to take care that the timeline is unchanged, but when I edit just audio, I manually trim parts when the edit is of different length. In this case “Ctrl+Shift+E” (Edit->Select->Select All After Edit Point) comes very handy.

As for limiters, the latest version of x42 plugins contain a peak limiter - x42-dpl - inspired by Fons Adriaensen’s Zita-dpl1 which always worked just fine for me.

Hmm, ripple edit has never given me any trouble in any of the DAWs where it is implemented correctly (Reaper, Magix products, Ardour etc). I do all my edits by ear to begin with in the sense that I create the in/out points while listening back. Any latency between me hearing and my finger moving is corrected as necessary in the crossfade editor. So no machine algorithms in play at all…

Is the x42 limiter true peak though?

No it isn’t. It’s a digital peak brickwall limiter, no upsampling.

Ardour supports true-peak and loudness normalization at export. That is generally much more useful that realtime dBTP limiting: allows to set a max loudness (EBU-R128) and max dBTP. Whatever limit is exceeded first is used to calculate the gain factor.

Check “Analyze Exported Audio” in the export-dialog to get a report.

Thanks, Robin! I’ll give this method a try…So I should use the x42 limiter with the usual -1dB ceiling and also set Ardour to normalize to -1dBTP?

Yes, that seems reasonable. The peak limiter on the master-bus can be used to trim some transients.
Those would otherwise attenuate the whole export when normalizing to (true)peak.

If you don’t have to deliver to broadcast industry, you could configure export-normalization to use

  • Loudness 0 LUFS
  • -1.0 dBTP

In this case the the loudness is not taken into account, because digital true-peak will always dominate the gain-factor.

Otherwise the established practice for program material is to aim for an average loudness of -23 LUFS but at most -1 dBTP (the defaults).

And here’s the companion volume to the first album:


Bach’s Inventions bring some warm memories from childhood. That used to be compulsory repertoire for anyone taking piano lessons.
Beautifully performed! Thank you for sharing.

tks for your share, c’mon up yeah yeah