HOA audio test featured r2

Hangouts On Air Studio Mode Audio Analysis

Here are my results from a series of tests run on Studio Mode audio in Google+ Hangouts On Air. The results come from the archived video on Youtube, not from within the Hangout itself.

Update: August 2014

What is the sound like in HD Hangouts?

Google upgraded Hangouts On Air to HD video a while back, and with that upgrade have come some improvements to the audio. This test is a bit more in depth than the original from April 2013, including audio files for you to hear the difference.

Testing

Playback of a high quality audio file of one of my original songs (a demo from 2012), first at average around -10db, then at full volume with average level around -2db. Provided is the original 320kbps MP3 file as a reference. At -10db you’ll be able to compare and contrast early Studio Mode (archived test from August 2012) with current Studio Mode. Also provided is a a test of Voice mode with the high quality filed played at -10db.

As another point of comparison, I’ve included audio from the same song clip played through Youtube Live.

Then I ran a series of frequecy sweeps through Hangouts On Air, going from 20hz to 20khz, at various levels of loudness: -3db, -5db, and -8db.

Spek is used to analyse the audio files and generate spectrograph images.

What’s new with HD Hangouts On Air?

The most obvious change is the upgrade to 720p video and some welcome updates to t20hz - 20khz sweep @ -8dbhe Hangouts interface. What we’re hearing on the audio side is an upgrade to 192kbps average bitrate AAC audio. This is an decent upgrade from the previous 96kbps AAC we had with the old 480p Hangouts.

The result of this, which you can see in the spectrograph images in the section below, is a reduction in noise in the 12kzh-15khz range. It also appears as though the 1khz notch filter is gone. The overall clarity of the sound has improved, but be wary of any claims that Hangouts On Air have “CD Quality” sound. They do not!

The audio meter in Hangouts is still poor and has in fact gotten worse when used with stereo input. The refresh rate of the meter is much faster, which is good for simple mono material, but with complex stereo material it flashes in a frenzied manner. One of my long standing complaints about Studio mode is that it needs its own proper 2 channel meter.

What hasn’t changed?

If you push the meters too high, you will pay for it. Look at the spectrograph images below and see how the distortion reduces when the average level goes from -3db, to -5db, to -8db. You do not have to get all the way up to 0db to get distortion.

The low-pass filter is still present, which removes any information above 15khz. Cutting out high frequency treble reduces bandwidth costs, and keep in mind that Google has to serve low bandwidth users as well. Eventually we may (don’t hold your breath) see more finite control over bandwidth settings, being able to choose specific bitrates for video and audio, but for the forseeable future this remains solely a feature of Youtube Live. These sorts of things would be very difficult to implement in a group video chat service, and possibly confusing for users (we can still dream).

Audio Samples

Now you get to hear the differences for yourself. Fortunately, I still have the video file for one of the earliest Studio Mode tests I did back in August 2012. Below you’ll hear the difference between the original high quality file, Youtube Live, early Studio mode from 2012, Studio mode from today, and Voice mode. The files have be arranged in order of quality, from highest to lowest. Also be sure to check out the spectrographs below to see what each streaming codec does to the audio.

Original File

“Get Up, Get Out” recorded by David Santy in 16bit/44.1khz .wav, transcoded to 320kbps LAME Mp3.

Youtube Live

Audio was streamed into Youtube Live via Wirecast for Youtube using the “Youtube Recommended” 720p output preset which averages 2500kbps. This was done with blank video to get the best audio I could, so video + audio may not quite sound as good. I’ll be working on a more in depth analysis of Youtube Live soon.

Go listen to the Studio Mode clips and come back to this one. Night and day! Hangouts On Air have a lot of growing up to do if they want to rival their older sibling Youtube Live. In addition to the lack of distortion, the extended frequency response, and the smoother low-pass filter, Youtube Live doesn’t clip as easlily as Hangouts On Air does.

Studio Mode @ full volume (2014)

Take note that while the original MP3 file has just enough headroom not to clip the meters (peaks at about -.08db), when played through a Hangout at full volume it clips rather frequently. It’s hard to say whether this is the fault of the Hangout audio, or the transcoding done for the archived video file. Either way, don’t run your sound this hot in a Hangout!

Studio Mode @ -10db (2014)

I said these were listed in order of quality, but this sample at a lower volume clearly has less distortion. This clip was provided as a comparison to the audio test from 2012 (below) and I wanted to put them next to one another.

Studio Mode @ -10db (from 2012)

This clip has a slightly different mix than the current one above, but you can still hear the difference in the distortion characteristics.

Today’s Studio Mode has less mud and greater clarity. Still not perfect by any means, but a decent improvement over what we had in 2012.

Voice Mode

Same file played through HOA but with Voice mode.

If you cringed while listening to this after hearing the original, you’re not alone. This is why you shouldn’t use Voice mode with high quality audio equipment, even if your show only features spoken word. It’s money down the drain. The thing you should know about this one is that I played the audio file into the Hangout at an average of -10db. Voice mode’s automatic gain made this one so loud that it peaked on several occassions. The noise gate cuts out all sound in the quiet bits.

Spectrographs

Now that you’ve heard the difference, see the difference here. In the spectrograph images from the “Get Up, Get Out” song clip, you wont be able to pick out the distortion from the image, but you can see how much of the audio each streaming codec is “chopping off” from the original file. The noise can bee seen in the spectrographs from the sine waves though. I didn’t include audio clips of the sine waves because, A) you could damage your speakers/headphones/hearing if you’re not careful and B) only lunatic audio nerds like me would get anything out of that. I’ll reconsider if someone requests it.

“Get Up, Get Out” Song Clip

I’ve arranged these in order of quality, from highest to lowest. As a point of comparison, I’ve also included one from a Youtube Live audio test (writeup on that coming soon).

Original Sound File

Get Up Get Out - Full Quality spectrograph

Youtube Live – 2500kbps stream

Get Up Get Out - Youtube Live Audio spectrograph

Studio Mode

Get Up Get Out-Hangouts On Air Studio Mode spectrograph

Voice Mode

Get Up Get Out-Hangouts On Air Voice Mode spectrograph

Sine Wave Sweeps in Studio Mode

20hz - 20khz sweep @ -3db

20hz - 20khz sweep @ -5db

20hz - 20khz sweep @ -8db

Test Video MediaInfo (Audio Only)

Audio
ID : 2
Format : AAC
Format/Info : Advanced Audio Codec
Format profile : LC
Codec ID : 40
Duration : 3mn 30s
Bit rate mode : Variable
Bit rate : 192 Kbps
Maximum bit rate : 203 Kbps
Channel count : 2 channels
Channel positions : Front: L R
Sampling rate : 44.1 KHz
Compression mode : Lossy
Stream size : 4.82 MiB (74%)
Title : IsoMedia File Produced by Google, 5-11-2011

Wrap Up

While Google has made some slight improvments to the audio quality in the past couple of years, we’re still more or less in the same position as we were then. Audio quality of Hangouts is nowhere near as good as that of video pre-recorded and uploaded to Youtube. I’m not expecting CD quality sound from a live streaming codec, but I think Google can get the sound in Hangouts a bit closer to that of Youtube Live. Unfortunately, there is a long way to go before we see that sort of quality.

You still have to be quite convervative with your gain going into a Hangout. Sparse arrangements with a single instrument and voice work quite well. The more density your mix has the poorer it will translate in an HOA.

Does this mean you shouldn’t attempt to do Hangouts with a full band or backing tracks? It can be done well with careful setup, but consider the following. While there are better sounding live streaming platforms out there (Youtube Live included), none of them offer the same sort of engagement that Hangouts On Air do. If you don’t need the face to face engagement, consider a proper live streaming service live Youtube Live instead.

Check out the conclusion section of the original audio test below for more pointers, and be sure to check out the Hangouts On Air Audio Setup Guide.

Thanks for reading!

 
 

(Original) April 2013 Audio Test

Run a 20hz to 20khz 2 channel frequency sweep through Hangouts On Air with Studio Mode audio codec enabled. Download .mp4 file from Youtube and run spectrograph analysis.

For reference, spectrograms comparing a 16bit/44.1khz lossless FLAC file sourced from an original studio recording and that of the archived video from the same file played through a Hangout On Air will be included.

Software
  • Reaper Digital Audio Workstation with “JS: Analysis gfx/spectrograph” plugin
  • SpekFree Acoustic Spectrum Analyser

Results

Bitrate

480p Youtube files have AAC audio tracks with an average bitrate of 96kbps which is fine for spoken content but far less desireable for music.

For comparison, the generally good sounding 720p Youtube videos have an audio bitrate of 192kbps.

Frequency Range

Low-pass filter* near 15khz. Dead air above that.

Noise

Lots of noise between 12khz and 15khz. Garbled sounding. You can see this noise in the spectrograph below, where the red/pink line grows very jagged as it approachs the 15khz low-pass filter cutoff. The noise isn’t going to be quite as obvious with with regular musical content, as you don’t usually see these frequecies at this sort of volume, but this noise can still contribute an overall “muddiness” to the sound.

The noise in the 12-15khz band makes it especially important not to push audio levels too high in Hangouts. Just because you don’t hear crackling associated with exceeding the 0db mark doesn’t mean there isn’t distortion.

Notch Filter

There may be a notch filter* at appoximately 1khz. Not sure whether this is a part of the Studio Mode codec or a result of the compression used to archive the video. Of the 3 frequency sweeps run, this drop-out occured in all three as the sweep reached 1khz. In the spectrograph (pictured below), this looks like a solid black line running top to bottom. It’s more easily seen in single sweep graph, but you can see where it repeats in all 3 sweeps in the graph of the whole test below.

Spectrographs

Frequency sweep test
Hangouts On Air Audio Test
One 20hz-20khz sweep
Hangouts On Air Audio Test
Whole frequency sweep test
Studio recording
Hangouts On Air Audio Test
Studio recording: Original

Hangouts On Air Audio Test
Studio recording: Archived video via HOA

Conclusion

Your mix is still the utmost

The most important thing to remember is that the distortion gets worse the louder your signal is, and you don’t have to exceed 0db to get distortion.

If your current method for setting levels is getting them just low enough to prevent clipping, your levels are still too hot.

That doesn’t just apply to Hangouts. While we’re dealing with a streaming audio codec here, even high quality analog-to-digital and digital-to-analog converters aren’t squeaky clean all the way up to 0db.

Give your mix a bit of breathing room!

Audio bitrate could be much better

Since the archived Youtube file is at “480p” quality with 96kbps audio track, the sound quality suffers. The worst of it is going to be heard with harmonically rich acoustic instruments like guitar, violin, or cymbals.

For music, I would like to see (uh, hear) an absolute minimum of 160kbps bitrate for the audio track. Finger’s crossed that Google is working on this presently.

I don’t have results from the audio inside a Hangout, as it’s rather difficult to do this testing without a second computer, but I can offer annecdotally that the audio sounds better in the Hangout than in the archived video.

It only goes up to 15khz?

A good pair of ears (read: young) can hear up to 20khz, but the low-pass filter isn’t really a big deal. It would be nice to see this go up to 18khz, but the 15khz cutoff isn’t as much of a problem as the audio bitrate.

What the low-pass filter at 15khz really means for the audio is that you lose a lot of the “air” from the sound, but to include that treble information would mean a much larger datastream. It’s a pretty fair tradeoff, especially for those dealing with limited internet speeds.

HOAs essentially have the same bandwidth, meaning the range of audio frequencies represented, as FM radio. Now that I’ve made that comparison, I should mention that it’s not the lack of treble extension that makes FM radio sound terrible. It’s the unholy amount of compression radio stations use to make their signal hotter.

How’s your internet connection?

Another point of consideration is that of bandwidth. For best audio/video quality in Hangouts a 4Mbps upload speed or greater is recommended, though anything over 2Mbps is serviceable depending on the consistancy of the connection. If you’ve got sufficient upload bandwidth, your download rate is probably fine since this is usually much higher than the upload rate.

Since internet service can fluctuate throughout the day, and by location, it is a good idea to test how good your connection is before upgrading your plan. You can test the quality of your internet connection, and whether you’re getting the speeds promised claimed by your ISP at SpeedTest.net and PingTest.net. If you’re not getting the promised speeds, or you’re seeing a lot of dropped packets, you’ll want to contact your ISP about it. If they don’t fix it, consider switching carriers.

Also, as I recommend in my Hangouts On Air audio setup guide, it is best to plug in directly to your modem/router via ethernet cable instead of using Wi-Fi whenever possible.

Check out the guide for the best Hangout sound

David Santy Google Plus Hangouts On Air Studio Mode Setup Guide

Test Video MediaInfo (Audio Only)

Audio
ID : 2
Format : AAC
Format/Info : Advanced Audio Codec
Format profile : LC
Codec ID : 40
Duration : 1mn 31s
Bit rate mode : Variable
Bit rate : 96.0 Kbps
Maximum bit rate : 102 Kbps
Channel(s) : 2 channels
Channel positions : Front: L R
Sampling rate : 44.1 KHz
Compression mode : Lossy
Stream size : 1.05 MiB (44%)
Title : IsoMedia File Produced by Google, 5-11-2011
Encoded date : UTC 2013-03-28 04:11:15
Tagged date : UTC 2013-03-28 04:11:16

Glossary

Low-pass filter: A low-pass filter is an electronic filter that passes low-frequency signals and attenuates (reduces the amplitude of) signals with frequencies higher than the cutoff frequency. More –>

Notch filter: A type of band-stop filter. In signal processing, a band-stop filter or band-rejection filter is a filter that passes most frequencies unaltered, but attenuates those in a specific range to very low levels. It is the opposite of a band-pass filter. A notch filter is a band-stop filter with a narrow stopband (high Q factor). More –>

2 comments

  1. Pingback: Replacing Skype with Google Hangouts - The Podcasters' Studio
  2. Ray Ortega

    Um…wow. Thanks man. I found your original article on setting up a Hangout On-Air Studio Mode a long time ago and was blown away. Today I was searching for anything that would give me some insights into what kind of audio I’m getting inside the HOA vs the archived YT file and was amazed to find this.

    You killed it. So good. Not sure this info exists anywhere else. While you couldn’t provide a spec for audio while inside the HOA, it’s good to know that it’s better than the archive which confirms what it sounds like to my ear.

    This means it’s better than 96kbps not compressed correct? I was looking for this info cause I do a whole show about podcasting as well as run a show based on G+ HOA (Podcasters’ Roundtable) and was going to talk about using Hangouts as a replacement for Skype. Also I’m working on ways to use it to record podcasts. The audio seems to be just as good IMO, at least when you are “live” (non-archived).

    96kbps would be great and if it’s better than that then the recommendation is a great one. I find Hangouts to be more steady/reliable than Skype these days. In addition to confirming the bitrate for live HOA, do you have any idea how it compares to specs you’d get from Skype?

    Thanks again for running this test. I look forward to sharing it.

    Best,

    Ray

Post a comment

You may use the following HTML:
<a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>