For the record

Mumble 1.2.3 recorder dialog

As you already know if you read the article on our Tracker Squash Meeting or listened to the buzz in #mumble the upcoming 1.2.3 release will have a recording feature. The work on this has progressed nicely and since we are now shipping the first snapshots with this new feature in them we think it is time to give you a proper introduction.

Let’s start with the basic rules we established for the recorder:

  • Recording is only possible while you are connected to a server and you can only record what you hear or say.
  • For privacy reasons we will not allow recording on pre 1.2.3 servers. (They are not able to properly warn their users)
  • There will be no ACLs to disallow it on newer servers. (We feel that if we restrict the feature that way the Open Source nature of our project will lead to forks with the warnings completely removed which is an even worse situation for the users)

Once you are on a server that fulfils these prerequisites you can open the Recorder via our Icon toolbar or the menu (Self->Recorder) and “Start” the recording. You can record for as long as your disk space lasts or until you grow tired of it. The data is encoded and written away on the fly. Supported audio formats at this time are:

  • wav (uncompressed)
  • ogg/vorbis (compressed)
  • au (uncompressed)
  • flac (lossless compressed)

You might ask: “Where is mp3 support?”. Well, mp3 has licensing issues which have prevented the library we use for encoding (libsndfile) from supporting it. As we do not want any legal troubles either we decided it would be best to keep it that way. We think as we offer multiple lossless formats this should not be much of an issue as you can always encode to mp3 afterwards.

Besides the format you can configure the recording mode. As a normal user you will most likely want to stick with the default: Mixdown. This means all audio streams Mumble receives are mixed together and stored in one audio file. This has the advantage of taking up little space and being easy to handle. But sometimes this is not enough: Say you want to record a Podcast interview or be able to cut out specific comments/noises from a single user in your recording later on. In this case Multichannel recording will be your mode of choice. In this mode Mumble creates a separate audio file for every user you hear. Once you are done recording you can then use audio editors like Audacity, Cubase or Ableton Live to put things back together and to do your editing. To make this as easy as possible Mumble adds silence to the beginning of the files of users who joined the conversation later and in general fills the gaps between talk with the right amount of silence to keep the tracks lined up.

Last but not least let me get to these funny %user, %date etc. things in the target directory and filename. These are so called template variables, if you ever did some (web-)programming you already know what those are good for but let me explain it for the rest. Template variables are special “words” that will be replaced with actual content by the program later on. Mumble currently offers the following variables:

  • %user – Inserts the users name
  • %date – Inserts the current date
  • %time – Inserts the current time
  • %host – Inserts the hostname

This means that if as your filename you choose something like “%time-%host” Mumble might expand this to the actual filename “11-23-21-mycoolserver.net.ogg” (Note: Mumble appends the right format extension on its own, it is not included in %host or anything). This helps you to customize the way your files are named the way you can handle them best later on. But as I said those variables can also be used in directory names. If you do a lot of recording all those files will clutter up your recording directory pretty fast and even with a great naming scheme you might have trouble to find what you want, this is especially true for multichannel recording with a lot of users. So if this is the way you will use recording you definitely want to add some template variables to your directory names. Mumble will automatically create these directories as needed so “C:\mygreatpath\to\somewhere\%date\%time\” will create a directory with the current date, in it a directory for the current time and in that will put your recordings. Do a second recording an hour later? Time has changed so it’ll end up in another directory. Same goes for the date of course.

This concludes this rundown of our new recording feature. For the audio geeks there is a small appendix with more information on what we actually save in terms of audio data. We hope you will test the hell out of this new feature before we release it with the 1.2.3 stable. Whether you find any bugs or not please give us feedback. The easiest way to do so is to leave comment below.

The Mumble Team

Appendix: So this is how we save it. But the audio geeks among you might ask themselves what we do actually save? Currently the answer to that is: Pretty much the same as you can hear minus positional audio (mono only at this time) plus what you say. Meaning it receives the same post-processing as the audio you hear does and if you have 44.1Khz audio output the file will contain 44.1Khz afterwards (this will change later, internally Mumble always operates at 48Khz and we want to retain that level of quality). The files contain either 24bit PCM (wav & flac) or the full 32bit float range (au & vorbis). Note that due to an internal API restriction our preprocessors currently only operate on 16bit PCM data. This is going to change soon which will make our whole processing pipeline 48Khz/32bit float. As far as codecs go the only lossy codec we offer is ogg/vorbis. For now we stick with the default quality level libsndfile chose for it which is 0.4.

This entry was posted in News, Tech talk and tagged , . Bookmark the permalink.

23 Responses to For the record

  1. Megagoth1702 says:

    I love you guys… :)

    Thank you!

    I really hope that more users would use that program… :-/

  2. Brett says:

    This is AMAZING. Thank you so much!

  3. Blu-Ray says:

    Thank you very much. I have been looking forward to this feature a very long time.

  4. Tommy says:

    Very very nice work, guys. Mumble is a great piece of software!

  5. Arthur Wiebe says:

    Great work!

  6. Niveks says:

    All that i wanted! thanks!

  7. sir says:

    Excellent. And you’re exactly right about forks without limitations. Open source! Thanks for doing it right.

  8. Himalde says:

    hmm…

    Nice first step but, imo the record feature isn’t finished before you can record yourself and others at the same time. Even if it’s only possible during multichannel recording. Until that’s possible this feature is useless for me. I know that it’s hard to do technically but FRAPS can record audio from two sources, even different sound-cards, at the same time so it is possible.

    No mp3 support is fine for me, FLAC support is just pure awesomeness :D

    Thank you so much for all your work. Cant wait for 1.23..

    • .D0T says:

      You are always included in your recordings. Sorry if I didn’t make that clear in the post.

      • kit says:

        You did: “Recording is only possible while you are connected to a server and you can only record what you hear or say.”

        He most likely got confused since it is such a short snippet on that, and such a long post on the multichannel uses.

        Love Mumble and am on a mission to convert all my gamer friends to it :) Keep up the great work!

        • Himalde says:

          I got hung up in:

          Recording is only possible while you are connected to a server and you can only record what you hear or say.

          I guess it was just me.

  9. Henne says:

    Nearly perfect.
    Very good job guys

  10. Pingback: DBCLAN.de » Neues Mumble Feature: Recording.

  11. hacki65 says:

    Nice new feature. Question: Is the warning that someone is recording seen in the overlay in-game ?

  12. RealNC says:

    Amazing! Thanks for all the hard work. This feature is not one of those one needs to use often, but *when* someone needs it, it’s an extremely useful one and saves a lot of frustration.

  13. Rav says:

    I’m not sure if I read it yesterday or not, is there record-possibility on serverside as well?
    We would use it for the raids, but having a central source for recording would prevent the raiders themselves risking to fill up their harddrives :)

  14. cR says:

    Is it in any way possible to record on a 1.2.2 server?

  15. Dynamic says:

    I know this is long after previous comments but…

    I’m thinking of the planned OPUS codec implementation in 1.2.4 (includes SILK, CELT and hybrid modes) and the fact that .opus file format (or stream format) is already getting some support (e.g. foobar2000 and Firefox, probably FFMPEG – usable for Audacity import)

    For the use case of podcast interview or conference-call recording and later editing (or even music jam sessions), it would be great to directly save the received data from each participant in the Opus format used to avoid an extra trancoding generational loss (to Ogg Vorbis) while saving disc space and processor overhead compared to lossless. Potentially, any delayed packets that arrive too late for live playback and require packet loss concealment during the live call, could also be saved, for improved quality in the final edit.

    In an ideal world, even, each user could save their own microphone signal in a lossless format (e.g. FLAC/WAV), ideally with some synchronisation or time stamping (or even just a low-latency sync signal, equivalent to a movie-industry clapperboard to start and end the recording processes simultaneously), so that the editor could, with the help of the other participants, assemble the podcast in lossless format after the interview/conference call/jam session to ensure no lossy transcoding and to allow manual per-participant volume-levelling/dynamic range compression/reverb/noise-reduction/stereo placement in Audacity or similar.

    • .D0T says:

      The snapshots already have the full Opus implementation, nothing left to “plan” there ;-)

      Recording directly to Opus is definitely desirable. It has some practical issues that would need to be worked out and it doesn’t fit that well with what we already have but it is definitely doable. Not sure when we’ll get to it though.

      The lossless additional mic recording sounds more like a job for a third party application though. The recording as is is already pretty powerful I think and we have to be careful not to overload the UI to much.

  16. treaki says:

    hi,

    how did you made this screenshot with alpha channel (transparency)? i would like to do so to…

    greetings treaki

  17. Jesus Meneses says:

    Your software is just fantastic!

  18. Pingback: Streaming Audio from Mumble using IceCast by ChinPen

Leave a Reply

Your email address will not be published. Required fields are marked *

*

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>