Bringing Audio into zynthian . . . . . and possibly recording it


Jofo mentioned a way of getting audio in in as a layer without using MOD-UI anyone remember what it is…?

I can’t find the mention of it…

I’m going to try out the new looper functionality … :smiley:


Simply create an “effect layer” and it will be autonnected to audio input is available :wink:

1 Like

Yep, that’s the way to do it… !

This is the clean guitar into audioinjector

The new meters really help get the tails correct ! :smiley:

And then add an effect . . .

A bit too much fast echo !

1 Like

It’s interesting to see how a little audio scavenging with a zynth begins to look using the everso simple crash record facility. . . .

Firstly the audio level meter makes this eminently viable !!!

And also the quck delete really helps keep things tidy.

so much thanks @riban :smiley:

As can be seen a little judious renaming a editing start to lead to a fairly good self maintaining list.

I feel the facilities I think are missing from this page:

1/ Play on zynthian
A play button that can tell the identified zynthian to play that wav file . . .
2/ Loop play on zynthian

3/ Record on zynthian… :smiley:

4/ Stop on zynthian.

5/ A refresh button , or better still a websockets driven auto refresh for list updates without going back through browser refresh or menu buttons.

And wildly . . .
I find that the process workflow involves a trip throu’ Audacity to trim the front and tail off.

Quite how one might implement that process in our simple process pipeline is open to MUCH discussion.
Auto detection always feel like you are just ridding for a fall, and the really only successful technique is display a waveform and then offer an interface that allows the selection of a zero crossing in and out…

Or alternatively perhaps its two encoders and rocking and rolling over the front and back.

Obviously this is no small feat, but I think it’s useful to consider how we might implement such things without actually setting finger to IDE
in our swiss army knife type role. . .

I worked with a VERY good VT editor ( 1000 Hz IS the right level for test tone, anything else is glorified meta data :- ) ). And he was constantly in record ALL the time.

Following that, Could we consider how we might implement a cache record and then mark the in and the point we actually set the file pointer to 0…?

IF we are considering wav files as proper currency in the zynthian world ( and we might well decide they aren’t. I know there is occasional talk of the DAW to bring it all together, but we have the record/play tool and I think it’s a really handy tool at this level. ) do we always trim to length ( overwrite or copy to new file?) or carry a ton of metadata around to describe levels, ins and outs and all manner of other possibly import stuff. . . ?

Questions, questions, questions . … :smiley:

1 Like

You forget the “download as mp3” button. Wav is not the best format for sharing on the forum …


.ogg is probably preferable, it was going to be the html 5 ‘standard’ audio format…
, or if you must options … :smiley:

I tend to use wav’s on the forum when I am confirming functionality. ogg get’s used for general wyleu mucking about and cat disposing . . .


OK! Let’s use Ogg :wink:

1 Like

Being able to manipulate / control the Zynth from the web interface may be a nice addition.

Although the elite sound engineers at the BBC know that 900Hz is right (so more pleasing on the ear) I will accept that the rest of the world (including the cloth-eared BBC TV lot) uses 1kHz test tones. (It is the EBU standard for example.) Other tones are available (including 250Hz, 440Hz and pulsed versions of these - oh those nights listening to 440 bongs keeping the transmitters on until we could send out test tones or those mornings, bongs similarly warming up all the transmitters before broadcast began…)

Moving on to @wyleu request to implement a full DAW within Zynthian :wink: - I am not enamoured with the idea of starting recording on first note. One may actually want some silence at the start of their recording (or some atmos or whatever) and it may prove challenging to configure the trigger level. It may be a useful (future) addition as an option. I have certainly used level triggered recording in the past to good effect.

Similarly, stopping recording when level falls below a threshold or on the last MIDI note played doesn’t sit right as a default. The user may want a longer or shorter decay / end of recording and the MIDI note thing is just plain wrong (does not account for decay of sounds).

A constantly recording mechanism which allows capture of ideas before you press the record button is wonderful but I worry it will be technically challenging with the RPi. It requires significant (volatile) memory / storage which is generally in short supply with the RPi. RAM is the ideal place for this but 1GB is already small for the core functionality of the Zynth. Disk cache is a poor second and we generally use flash storage which hates being written to. We would likely kill any such flash storage very quickly. I love the idea and would like to hear any ideas of how we might implement it avoiding these issues.

The ability to “top and tail” a piece is a nice feature that (along with normalisation) is probably the most frequently used workflow - indeed many users may only use an audio editor for this purpose. Adding such a feature to the audio recorder / player (/editor) could prove popular but remember, this is a Zynthian, not (yet) a ZigitalZudioZorkstatian. The Swiss Army Knife tag is nice but my last knife fitted in my pocket, not on a truck I had to drag behind me to camp.

Regarding an Edit Decision List - that does feel like a step too far! Non-destructive editing is a real boon that digital technology brought to the audio world but I am sceptical about what benefit it will offer us. (I could be wrong - it has been known!!!) I would say that, if we want to add some (simple) editing features we should start with actual changes to the files and leave the clever stuff to our DAWs. I like the idea of being able to record something within the Zynth real-quick-like then use that recording in my DAW as the basis of something bigger or a contribution to a project. I don’t anticipate using the Zynth as my DAW any time soon. (I know some workstations like the Korg beast allow this and maybe that is an area we might explore at some point.)

I have written multitrack recorders for the RPi but have encountered issues with bandwidth limitation so they tend to sit in the pile of projects for a rainy day. (I know we get a lot of rain here in the UK - more where @wyleu lives - I live on the “Sunshine Coast”) but the pile is oh-so high.)

Lots to think about - let’s continue our discussion and avoid scoping the unachievable or the undesirable (both attributes which often become evident when it is too late).


I think, the Zynthian should be an instrument and the next big feature I like to see is a better looper integration. Esp. for drums.
I want to trigger some loops on my pads, play along, record new ones and assign them to pads on the fly while playing.
And we should be able to share the clock in a zynthian network, so that we can play with several people in a zynthian mesh-up.


Having just listened to Dad’s Army on Radio sev … four extra. I can concur that the radio medium is a poor absorber of such rich fodder as the Moving Image can provide, but, moving on :-D.

880 ‘A 440’ 76 71 -10dBFS SNAPSHOT REF to ZynSineX.sf2 Zyn Sine A 440
884 A 442 76 71 -10dBFS SNAPSHOT REF to ZynOboe.sf2 Zyn Oboe A 442
1000 EBU ref 83 71 -10dBFS SNAPSHOT REF to ZynOboe.sf2 Zyn Sine EDU

As long as we can guarantee the existence of the sample files, we should be ok.
Indeed that would act as a pretty good confirmation that we are talking to a generalised zynthian filesystem. Once we’ve made that assumption things get a little bit more simple.

The silence on the front divide can, to an extent be handed off to PolyPhone, but to pass through that chain you need to export/import.

I’d be interested to see how an auto detect on start would fit. I suspect that almost turns it into a looper at little or no cost.
The tail is Really involved in the loop/ not loop scenarios’

I don’t know what jack load MIDI recording requires, but I am growing increasingly fond of the idea of Recording MIDI as a matter of course when Audio is recorded. Could a MIDI stream be economically ( CPU & Disk space) grabbed at the same time?
From one use case I’d think of cued theatre (gun shot, creaking door etc) use you would probably want an instant start.
Even pre atmos’d effects playin would be precisely cued, it’s probably the thing that might be said to ALWAYS start the value add… :smiley:

IF we do we have an interesting source of meta data, perhaps we could start an zynth unofficial protocol of MIDI Note Event Off at start and end of sample. ?

IT might even save a few panics… :smiley:

If the auto audio start detection fires off the MIDI Event Key and then writes the Key off then we can present a choice of from start of recording/first sound detected. Which might allow us to keep it within GUI until you get to the preview/delete to send off to the outside world for renaming and misuse…