Ferrum WANDLA - Impressions Thread

[GoldenSound]

Just to update on the GoldenSound Edition conversion/upgrade programme!

Development is now completely finished. The latest firmware update and Ferrum Control App update both have all the necessary stuff for it. It'll be available to purchase on Aug 21st, as Ferrum will be on holiday for a couple weeks and it would not be a good idea to launch whilst they're away in case people need assistance.

If you were looking to upgrade your wandla, Aug 21st is the date!

1 week to go!

 

[Ferral]

1 week to go!

Does removing the MQA capabilities have any benefits?

 

[GoldenSound]

Does removing the MQA capabilities have any benefits?

Not in terms of SQ. Although on other devices with mqa/non-mqa variants that may not be the case as many of them do oversampling differently/worse on the MQA version but not the non-mqa one. The Wandla only used the MQA filter when actually playing MQA though.

MQA was removed because we needed all of the available compute power for the spatial enhancement. Running spatial enhancement wasnt possible whilst also running the MQA core decoder

 

[Ferral]

Not in terms of SQ. Although on other devices with mqa/non-mqa variants that may not be the case as many of them do oversampling differently/worse on the MQA version but not the non-mqa one. The Wandla only used the MQA filter when actually playing MQA though.

MQA was removed because we needed all of the available compute power for the spatial enhancement. Running spatial enhancement wasnt possible whilst also running the MQA core decoder

you mentioned something about more headroom I do not understand. Is there more headroom in the GSE? please pardon me, as I am actually a pure 2ch listener, and I'd like to know the 2ch differences before I pull the trigger. most likely I would support you though. excellent work!

 

[OhmsClaw]

you mentioned something about more headroom I do not understand. Is there more headroom in the GSE? please pardon me, as I am actually a pure 2ch listener, and I'd like to know the 2ch differences before I pull the trigger. most likely I would support you though. excellent work!

I imagine more headroom refers to freeing up processing power to prevent intersample overs and retain the "overhead" during processing so that headroom is achievable when you turn the volume up for the pre-out or choose fixed regardless.

Edit: in addition to DSP on the GSE

 

[GoldenSound]

you mentioned something about more headroom I do not understand. Is there more headroom in the GSE? please pardon me, as I am actually a pure 2ch listener, and I'd like to know the 2ch differences before I pull the trigger. most likely I would support you though. excellent work!

Yeah the GSE has more digital headroom to account for larger intersample overs.

The standard Wandla has some already so can account for small/mild ones. The GSE is effectively immune to intersample overs entirely


All the features work for 2ch as well as headphones

I've got a 2ch system as well so it was important for them to work with both.

The impact+ and tube modes work the same and can be used on either headphones or 2ch.

The spatial enhancement has a speaker mode and a headphone mode

 

[axle_69]

I imagine more headroom refers to freeing up processing power to prevent intersample overs and retain the "overhead" during processing so that headroom is achievable when you turn the volume up for the pre-out or choose fixed regardless.

Edit: in addition to DSP on the GSE

If I'm interpreting this correctly, if the original recording is 16 bits the integer amplitudes vary from -32768 to 32767. A sample at maximum level has a value of 32767. When processing at 32 bits we would multiply by 2^16 to scale to the new range. The next step is to oversample/upsample/interpolate, if we have two samples at maximum level that are the crest of a wave and we want to create samples in the middle, the samples in the middle can't have higher value than the maximum and the crest of the wave looks truncated. We need around 3 dB between the maximum of our signal and the maximum that is allowed (0 dB), therefore we cannot scale by 2^16, we need to scale by a lower value to give room for the extra samples between our 2 samples at the maximum to follow the nice curve of a crest. The same goes for the minimum. This is only a problem for very loud signals.

 

[GoldenSound]

If I'm interpreting this correctly, if the original recording is 16 bits the integer amplitudes vary from -32768 to 32767. A sample at maximum level has a value of 32767. When processing at 32 bits we would multiply by 2^16 to scale to the new range. The next step is to oversample/upsample/interpolate, if we have two samples at maximum level that are the crest of a wave and we want to create samples in the middle, the samples in the middle can't have higher value than the maximum and the crest of the wave looks truncated. We need around 3 dB between the maximum of our signal and the maximum that is allowed (0 dB), therefore we cannot scale by 2^16, we need to scale by a lower value to give room for the extra samples between our 2 samples at the maximum to follow the nice curve of a crest. The same goes for the minimum. This is only a problem for very loud signals.

Mostly yeah. Though the 3dB thing is just a general 'most stuff will be fine' rule. Actually intersample overs can be arbitrarily high, there was an interesting paper/technnote on that somewhere but I can't for the life of me find it. Essentially they showed that with a synthetic signal you can create much higher intersample overs.
The 16 vs 32 bit thing is not actually the issue with intersample overs. If it occurs at one bit depth it'll occur at others since max value is still max value even when described more precisely.
In the Wandla GSE the mathematical precision for features like the Impact + mode are actually done at 64 bit to ensure maximum possible quality

In actual music though, intersample overs do occur often. A few examples:

• Travis Scott and Drake’s “Sicko Mode” (2.4 dBTP)
• Dua Lipa’s “Levitating” (1.8 dBTP)
• Doja Cat’s “Say So” (0.8 dBTP)
• Carly Rae Jepsen’s “Call Me Maybe” (0.8 dBTP)
• Mariah Carey and Boyz II Men’s “One Sweet Day” (0.7 dBTP)
• Bruno Mars’ “Uptown Funk” (0.4 dBTP)

Strong transients in music or music where there is significant bass content, therefore leaving less available room/crest factor for the higher content, will also cause more frequent clipping.
When using a minimum phase upsampling filter the true peak level may often be higher too, and there is some music which I've seen with slightly over 3dB intersample overs.

The Wandla GSE can handle intersample overs in excess of 3dB and this extra headroom also means that features like spatial enhancement or Impact + can be enabled without requiring any reduction in volume/output level

 

[axle_69]

16 bit was the example of an original recording and 32 the processing side, but yes the issue is the same independently of the bit depth of the recording as long as it reaches the maximum value or close.

 

[SolarCetacean]

In actual music though, intersample overs do occur often. A few examples:

• Travis Scott and Drake’s “Sicko Mode” (2.4 dBTP)
• Dua Lipa’s “Levitating” (1.8 dBTP)
• Doja Cat’s “Say So” (0.8 dBTP)
• Carly Rae Jepsen’s “Call Me Maybe” (0.8 dBTP)
• Mariah Carey and Boyz II Men’s “One Sweet Day” (0.7 dBTP)
• Bruno Mars’ “Uptown Funk” (0.4 dBTP)

Strong transients in music or music where there is significant bass content, therefore leaving less available room/crest factor for the higher content, will also cause more frequent clipping.

What software do you use to measure these intersample overs? The biggest intersample overs I think I've encountered were in Excision & Sarah de Warren's "Back to Life". Based on PEACE gui's clipping meter, I had to reduce digital volume by nearly 6 dB before the clipping stopped. But I never did an actual analysis of the file. Though I use an AAC encoded file, and I've found that lossy encoding often introduces or adds additional intersample overs.

 

[GoldenSound]

What software do you use to measure these intersample overs? The biggest intersample overs I think I've encountered were in Excision & Sarah de Warren's "Back to Life". Based on PEACE gui's clipping meter, I had to reduce digital volume by nearly 6 dB before the clipping stopped. But I never did an actual analysis of the file. Though I use an AAC encoded file, and I've found that lossy encoding often introduces or adds additional intersample overs.

General process to check is to apply -6dB attenuation to a track with something like adobe audition, then upsample using either Audition itself or something like HQP. Then look at the track and see where the highest amplitude is and add 6dB.

So if after doing the above you have content at -2dB, adding the 6dB back to counter the digital attenuation we applied earlier would mean that's a +4dB intersample over

 

[Ferral]

Is anyone using Roon->HQplayer->Wandla? Can we share what are the best settings to use and do we convert DSD files to PCM, as it seems the Wandla performs better with PCM. Thanks in advance!

 

[rdbogdan]

1 week to go!

Sales are live for converting plugin on Ferrum site in the shop section, bought it an hour ago, hopefully i will have some time in the evening for some listening.

 

[akukula]

Sales are live for converting plugin on Ferrum site in the shop section, bought it an hour ago, hopefully i will have some time in the evening for some listening.

Pls do give us a heads up if its worth it. Much appreciated. Thanks!

 

[krude]

Pls do give us a heads up if its worth it. Much appreciated. Thanks!

I did the update. Is it worth 600 euro? It depends.

So far I tested on Utopia. The effects are subtle, but noticable.

If you have a very well sorted out system and very well recorded music it will make it worse.

But if you have music that would benefit from a bit more bass and body and a bit wider stage, then it works as expected.

One use case for me is rock / metal. A lot of albums are recorded really lean and adding that little extra body is a welcome change.

If I try to add more bass to a very good electronic bass heavy mix it makes it muddy and looses clarity and definition.

But ... it's good to have options right?