Jump to content
  • mitchco

    Objective: Audeze LCD-4z Headphone Review With Measurements

    image1.jpgIn this objective review of the Audeze LCD-4z headphones, I measure the frequency response and provide binaural recordings of the headphones so that folks can download and listen to music samples. The music samples are compared to the original track so one can hear, relatively speaking, any differences between the source and headphone reproduction.


    In addition, there is a comparison to the NAD Viso HP50 headphones both from a frequency response and binaural recording perspective. The binaural music recording switches back and forth every 10 seconds between the two headphones, so one can evaluate tonal differences between the two.





    A few points to put this review into context. Unlike most headphone measurement rigs that use a dummy head, I am using my own noggin and ears as the dummy head (still a dummy head says my wife). This way I can adjust the fit and feel of the headphones to sound just right to my ears. Many headphone measurement rigs require a “compensation curve” to interpret the raw headphone measurements. Tyll does a good job of explaining that in the link above. However, in this particular case, I have found a set of in-ear measurement mic’s that seem to not require (much of) a compensation curve, up to 10 kHz anyway. More on that point later.


    Just like loudspeakers, there are subjective listening tests correlating to objective measurements to indicate that there is a “target curve” for headphones that correlates to what sounds neutral or preferred or accurate. Sometimes referred to as the Harman target curve. Tyll, in the link above, discusses it and I wrote a summary, “The Science of Preferred Frequency Response for Headphones and Loudspeakers” peer reviewed by Sean Olive


    There is an educational presentation by Sean and team on, “The Perception and Measurement of Headphone Sound Quality: Do Listeners Agree on What Makes A Headphone Sound Good?” It is interesting to read about the controlled subjective listening tests, objective measurements and conclusions. Since then, the subjective listening tests have been repeated several times, with hundreds of participants, arriving at the same objective results and encapsulated in a more recent 132 page presentation.


    Regardless of headphone compensation curve, I am using the NAD Viso HP50 headphones as a comparison point with the LCD-4z’s. The reason why HP50 is because they sound and measure the most neutral compared to a reasonably large sample size of headphones. From the link above, Tyll’s comment about the HP50’s (and Focal Spirit Professional), “These two headphones are among the most neutral I've heard, and they do match the Harman target response quite well relative to other headphones I've measured.” At the time Tyll has measured and reviewed some 200 headphones. Having measured and reviewed the HP50 myself, I achieved similar measurement results and agree with Tyll’s subjective statements, quite a neutral response, albeit down in level in the top octave.


    Using my measurement setup, whatever the frequency response curve measures out for the HP50 can be considered, relatively speaking, neutral or a point of reference for comparison. Then by measuring the LCD-4z, we can compare to “neutral”. This will correlate with the binaural recordings as well. Meaning, with the binaural comparisons of the Audeze LCD-4z to the source track and to the NAD HP50 headphones (that measure and sound neutral), we can characterize the tonal response of the LCD 4z’s relative to an ideal or neutral reference.


    On a technical note, the Audeze LCD-4z headphones have a rated impedance of 15 ohms. This means many sources can drive these headphones to satisfying levels without the need for a separate headphone amplifier. However, with this low of an impedance, one needs to be careful that the source device driving the headphones should have, as a rule of thumb, an output impedance at least 1/8th the impedance of the headphones. So 1/8th of 15 ohms, is less than 2 ohms. Therefore, the output impedance of the device used to drive these headphones should be less than 2 ohms. My Lynx Hilo headphone’s output impedance measures about 0.08 ohms, so no problems. Here is a link to several headphone devices output impedance that one can compare.


    At the end of the article I include an appendix on the calibration and verification of the test gear that I used to measure frequency response and make the binaural recordings. In a nutshell, the test equipment has flat frequency and phase response with low distortion. I.e. a transparent recording chain. With the preamble out of the way, let’s move on to the objective measurements.



    Objective Measurements


    If you don’t want to look at graphs and would rather listen to the binaural recordings, click here.


    Here is my methodology to reduce variation in getting a good headphone seal and consistent frequency response measurements. I took ten measures or five stereo pairs, and for each measurement, I adjusted the binaural mics and the headphones to give the best and most consistent frequency response. 


    Here is the left and right channel frequency response for one pair of measurements using REW:





    As one can see, an incredibly flat frequency response from 20 Hz to 1.6 kHz, with a -3 to -4 dB dip between 1.6 kHz and 4.1 kHz. Then a +5 to +8 dB peak at 5.3 kHz. From there, a -20 dB dip at 7.8 kHz and then a +16 to +18 dB peak at 9.9 kHz. Then another -20 dB dip at 14 kHz. Getting a good seal with these headphones is no problem, but a bit of a bumpy ride in the treble.


    Now some of that is due to timbral issues with head related transfer function (HRTF) and that it is my head. It is also related to binaural recording techniques whether using the blocked ear method or microphones at the ear drums. Further, getting the binaural mics into one’s ears in the same place each time has variability, as does the fit and positon of the headphones. In the case above, even variability at the top end between the left and right channels due to the right mic’s position.


    It isn’t easy trying to reduce these variables to record realistic headphone measurements that are repeatable. But close enough to be the same trend in tonal response to match, which is what we are after. One can also compare to other measurements, as we will with the NAD Viso HP50 and correlate the differences, not only in the measurements, but in the binaural recordings as well.


    Let’s look at a distortion measurement. I would not put too much emphasis on the absolute distortion levels as the test was performed at regular listening level, not to a measurement spec. Also, I have no idea how much distortion is being contributed by the microphones as there is no distortion spec for them. What we are interested in is the overall spectral shape of the distortion versus frequency to get an idea of where there might be some potentially audible issues. Rising distortion in the low end is typical of any headphone or loudspeaker for that matter, often in large numbers, but reasonably low for the LCD-4z:





    There is a rise in distortion, relatively speaking, from 3.5 kHz to 8.5 kHz. For those wondering why the distortion measures only goes up to half the sample rate can read the REW help.


    Let’s bring in the NAD HP50’s so that we can make a measurement comparison. I matched the sound pressure level for both headphones by adjusting volume using pink noise as the source through REW so that they matched in level. This was both for the measurements and binaural recordings. I followed the same procedure as mentioned at the top of the objective measurements. HP50 left and right channels – no smoothing:






    Here we can see an extended response down to below 20 Hz and mostly flat out to 1 kHz with a bit of dip from 1 kHz to 3.5 kHz and the then a rise and only after 10 kHz we see a roll off with some peaking and dipping. Note it is much smoother in the treble than the LCD-4z relative to the magnitude of the peaks and dips.


    The one thing to notice is that I could not get the best seal on the right ear cup and shows as a channel imbalance in the low end. Don’t really notice it when listening to music. The HP50’s have small ear cups and for my large ears, with glasses and long hair, make it next to impossible to get a good seal. They did at the beginning, but that was due to the brain crushing clamping force applied until I nearly broke the darn things stretching them out.

    Again, a little bit of right channel imbalance at the top as a measurement artifact due to mic positioning. But comparing to my previous measurements of the HP50, I can verify consistency and repeatability:






    The red and blue traces are from this review and the green and gold traces are from my previous review of the HP50’s from 1 ½ years ago. As we can see in the previous review, aside from the small low end seal differences, the left and right channels track closely to about 10 kHz. This is why I know that the blue curve in this review is with the right mic not quite positioned correctly. But the left earphone from this review tracks very well with both left and right channels from the previous measurements. As can be seen, I have a little work to do on getting both mics in the same place for left and right measurement consistency.


    A distortion plot:






    Pretty smooth with rising distortion in the low end, which our ears are insensitive to, a narrow spike at 1.93 kHz and some rising distortion between 4 kHz and 6 kHz with the right earphone a little higher than the left.


    Now let’s look at some comparisons so we can make predications on what we think we are going to hear in the binaural recordings.


    Here I have overlaid the LCD-4z and the HP50’s frequency response – both channels, no smoothing:






    Both headphones, from a tonal perspective are very close from 20 Hz to 700 Hz, with the HP50’s having a bit more output in that range. The HP50’s are up to -3 dB down from 1 to 1.5 kHz and then it is a bit of a mess to try and sort out after that. To make it easier, I have overlaid the individual left and right frequency responses for comparison. First the left side:






    As previously mentioned with the HP50 (green trace) having a bit more low and mid frequency output, both have a pretty smooth response up to 6.5 kHz. We see the LCD-4z (blue trace) have a -3 dB drop from 1.5 kHz to 4.5 kHz relative to the HP50’s. There is a -12 dB drop at 7 kHz relative to the HP50’s. Conversely, we see about a +10 dB spike at 9.3 kHz for the LCD-4z relative to the HP50. At the very top end, we see that the LCD-4z has +22 dB more output that 18.3 kHz than the HP50’s.


    Looking at the right earphone:






    Good tonal match out to 6.3 kHz. Relative to the HP50’s there is about -10 dB drop in the LCD-4z at 7.6 kHz. At 10 kHz, the LCD-4z have about +10 dB more output relative to the HP50’s at 11.6 kHz and the LCD-4z’s have about +16 dB more output than the HP50’s at 17.8 kHz.


    Note these graphs are not smoothed. Meaning they are full resolution measurements, but it is commonly acknowledged that our ears don’t hear that narrow bandwidth of frequency resolution. Our ears hear about 1/6 to 1/12 octave bandwidth. So if I apply 1/6 octave smoothing, we can get a better handle on the tonal differences that we expect to hear relative to the two headsets as compared to each other and to the original source music track.


    Rather than repeating the same graphs above with the smoothing, let’s just look at the left channel between the two headphones. I feel this best represents each headphones sonic signature from a measurement perspective:






    • The HP50 (green trace) have about 4 dB more output in the low frequencies and lowers mids than the LCD-4 from 20 Hz to 750 Hz.
    • Relative to the HP50, the LCD-4z has up to 2 dB more output from 800 Hz to 1.5 kHz.
    • The LCD-4z’s have about -3 to -5 dB less level from 1.5 kHz to 4 kHz as compared to the HP50’s.
    • LCD-4z are -7 dB down at 7.4 kHz relative to the HP50.
    • LCD-4z +9 dB at 9 kHz.
    • LCD-4z +8 dB at 15.4 kHz
    • LCD-4z +21 dB at 18.7 kHz.


    Comparatively speaking, I expect the HP50’s to have a bit more bass and lower mid output than the LCD-4z’s, both exhibiting a nice smooth response (LCD-4z ruler flat!) with extension to below 20 Hz. Might not be much of an audible difference in the 800 Hz to 1.5 kHz range, but could make the voice on the HP50’s not stand out as much in the midrange. The LCD-4z’s will likely sound recessed in the 1.5 kHz to 4 kHz range, which is also the range our hearing is the most sensitive. So not as bright sounding in the upper mids as compared to the HP50’s. That plus the LCD-4z with the peaks and valleys in the treble response, and the LCD-4z’s having (much) more high frequency output than the HP50’s past 13 kHz will be the most audible differences, I predict.


    Distortion plot of both left channels:





    The HP50’s have a bit more distortion in the low frequencies and the LCD-4z more distortion in the 4 kHz to 11 kHz range. Seems to be tied more to the magnitude response level differences... I am not convinced at what level our ears can hear THD as they are frequency dependent. It also requires calibration with a set of mics that have known distortion specs, which these binaural mics do not. So how much distortion is from the mics versus how much from the headphones? And how much is really audible? Not sure if this is providing value as our ears seem to be (far) more sensitive to frequency response variations than anything else. Want to hear it for yourself? Try ABX Testing and Distortion.


    Let’s listen to see if these frequency response differences are audible in the binaural recordings.



    Subjective Listening using Binaural Recordings


    OK here is where the fun begins. I recorded a snippet from Tracey Chapman’s Fast Car. It is a familiar recording with good bass and dynamic range with her voice mixed up in level so it stands out and makes it easy to hear any tonal imbalances. Also pay attention to the “s” in her voice as it can range from dull sounding to overly sibilant. I am using a portion of the song on the basis of fair use for the purpose of evaluation and research. I am not doing this for financial gain. Please delete these music samples once you are done with the evaluation.


    I level matched the binaural recording to the original track in my DAW (Mixcraft) and then using the DAW’s automation, synchronize the two stereo tracks and switch back and forth every ten seconds so you are getting the benefit of instant switching between the source music track and the binaural recorded track. Looks like this in Mixcraft:





    There are some “left in” edits when the tracks switch. These can be used as an audible cue to focus in on the new sound to assess the differences before you get used to it as we quickly adapt.


    I picked a random spot, just before Tracy sings and then a couple of choruses. We start off with the original track for ten seconds and then switch to the binaural track of the Audeze LCD-4z, then back to the original, switching back and forth every ten seconds until the end of the music selection. 



    notes.pngTracy Chapman Fast Car Audeze LCD-4z (WAV File - 33.1 MB) (ZIP File - 29.9 MB)



    To my ears, the bass and mids sound pretty close together, but the upper midrange and treble sound different to my ears. The tonal response of the LCD-4z sound like their measured frequency response where some frequencies are missing in Tracey’s voice, and other frequencies are too much, giving it a distinctive tone and a little too bright sounding overall in the top octave. Overly sibilant sounding as compared to the source is what I hear. While this may be some folks preference, not that there is anything wrong with that, it does deviate from a neutral response like the ruler flat bass and mids that the LCD 4z’s do have.


    There is some context here as well. The headphones are recorded using my ears with a certain HRTF and is going to have “some” impact. But as you listen, when it is a match, like in the bass and mids, the source and the binaural recording sound amazingly similar.


    I would love to hear your feedback on what you hear. Are you hearing similar to what I am hearing or different?


    Now let’s take a listen to the source track and compare with the NAD Viso HP50. Again, starts with the source track and then in ten seconds switches to the HP50, then ten seconds later, back to the source, switching back and forth every ten seconds to the end of the music selection:


    notes.pngTracy Chapman Fast Car NAD HP50 (WAV File - 32.6 MB) (ZIP File - 29.3 MB)


    To my ears, again the bass and mids sound very similar to the source. Also, Tracy’s voice sounds similar to the source from a tonal perspective. It does not sound “filtered.” However, as seen in the frequency response measurement, the top octave after 10 kHz drops off fairly quickly as the attack on the guitar strings, the hi-hat/cymbals and Tracy’s “s” all sound down in level (i.e. dull) compared to the source track. A little shelving eq or treble control boost above 10 kHz would bring the level back up. Then it would be more difficult to tell the two apart.


    There are balance issues between left and right channels that one can hear in this binaural recoding (and the LCD-4z too) as the center image shifts a bit when switching. This is due to the binaural mics not sitting perfectly in each ear canal.


    What are you hearing?


    Finally, a direct comparison of the LCD-4z to the HP50’s: 


    notes.pngTracy Chapman Fast Car LCD4z and HP50 (WAV File - 32.9 MB) (ZIP File - 29.5 MB)



    Regardless if listening over headphones (preferred) or speakers, and regardless of individual HRTF’s, it is a direct comparison where one can hear the “relative” differences. The two big differences that stand out for me is that Tracy’s voice sounds close to as it should with the HP50’s whereas listening to the LCD-4z’s sounds like the upper midrange of the frequency range is missing in her voice and then accentuated too much in the treble (listen to the “s’s”). The treble response in the HP50 is rolled off, but easy to fix with a basic high frequency (shelving) tone control. Not so easy with the LCD-4z as it is going to require parametric eq to deal with the ups and downs in treble frequency response.


    What are you hearing as the differences between the two headsets?





    The Audeze LCD-4z have an incredible flat frequency response in the bass, with good extension below 20 Hz and flat midrange response up to 1.5 kHz. This is evidenced by the measurements and correlated with the binaural recording comparing to the source track. I mean it sounds and measured perfectly flat with excellent matching channel balance. Also matches the Harman target curve for a neutral frequency response pretty well up to 1.5 kHz.


    From there, we get into deviations from neutral with a little lower level output in the 2 to 4 kHz range, which is right in our human ears most sensitive range. Then we get into some peaks and dips in the upper midrange and treble. Not only measured, but also heard on the binaural recordings, both comparing to the original source track and the NAD HP50’s. Finally, overall a little too much high frequency energy past 13 kHz, both measured and heard.


    Of course there are complicating factors related to measuring headphones and making binaural recordings using a person’s (or dummy) ears (mics fitting properly and repeatedly is one example). While it may not represent the absolute truth, from a comparative perspective, it appears to be quite accurate as there is not only good correlation between the measurements, but also comparing to the source music track. 


    The direct comparison differences between the two headphones should hold true, regardless of one’s own HRTF, headphone or speaker brand while listening to the binaural recording. As a side note, I believe this binaural recording technique is sensitive enough and enough resolution that if there is an audible difference when directly comparing two devices, using music as the test signal, then the recording will pick it up and accurately represent it on playback.


    One can always eq the LCD-4z’s towards a neutral response if desired. The data to implement parametric eq (PEQ) is contained in this review where inverse filters for the peaks and dips could be created for both channels. Bring up the response in the 1.5 kHz to 4 kHz range and then bring the overall treble response down to neutral (or taste) and. I would use the 1/6th octave smoothing graphs as the guide for PEQ. Or if you really want to fully optimize the response, one could get a pair of binaural mics, a mic preamp and use DSP software like Audiolense or Acourate to accurately tailor the response to be neutral, and matched to your own HRTF. Or even try the Impulcifer which is like an open source version of the Smyth Realiser.


    I did try Audeze Reveal, which has an eq preset for the LCD-4z, as a plugin in JRiver. To my ears, it was a subtle difference and did not tonally change or reduce the treble response. I did not get a chance to see how customizable Reveal is.


    Finally, make sure the device you are using to drive the LCD-4z follows the 1/8th rule for impedance matching. For a 15 ohm headset, the driving device needs to have less than 2 ohms output impedance.


    I hope you are enjoying the music!



    Appendix: Calibration and Verification of Measurement Equipment


    The purpose of this section is to list the test equipment I used and verification measurements performed to ensure a transparent signal path so as not to influence the headphone measurements or binaural recordings of the headphones. Test sample rate is 48 kHz and all digital and analog levels on the Lynx Hilo are set to 0 dB unless otherwise noted. 


    Sound Professionals MS-TFB-2 Binaural Microphone


    I have tried many binaural microphones and dummy head set ups over the years. I find the Sound Professionals MS-TFB-2 to be a unique set of binaural mics as when put in-ear, they have an incredibly flat frequency response. No other binaural mics I have tried have this characteristic. Mine were ordered with the XLR ends and phantom power converter so they can be used with a mic preamp that has phantom power.





    You can read the rest of the specs from the link above. Contacting the supplier who indicates the frequency response was measured in an anechoic chamber using the standard for measuring microphones. While I can’t verify the frequency response, I can say over the years using this mic set for a number of binaural recordings of music and nature events plus audio gear, I am reasonably confident of the spec. A shout out to Chris at The Sound Professionals for answering my many questions in a timely manner.


    ART PRO MPA II Microphone Preamplifier


    The ART PRO MPA II is a stereo microphone tube preamplifier that in a shoot-out out ranked several top name brands. Using REW, I routed the digital test signal through the Lynx Hilo DAC analog output and input to the mic preamp with the mic preamp’s output feeding into the Lynx Hilo’s line input ADC and routed back to REW for display. For using a studio pre as a measurement preamp, it did pretty well:






    -1 dB down at 15 Hz and ruler flat to over 20 kHz using a 48 kHz sample rate. I did run a frequency response test at 192 kHz sample rate and the preamps response was -1 dB down at 50 kHz. Looking at Phase:





    Flat phase response.





    Ah, we see the right channel with a higher level of distortion than the left. The good news is that it is linear across the frequency range and at 0.02 percent distortion, very likely inaudible. Btw, that’s with 20 dB of microphone preamp gain applied. What causes the small channel imbalance? It’s just tubes man. I should also point out, these are the same settings (i.e. 20 dB of preamp gain) when making the binaural recordings. The binaural recordings may not be the most quiet, not only due to my environment, but I was quite conservative with headroom. I could have achieved another 10 dB or even 20 dB at best signal to noise ratio, but in the context of “comparative recordings”, this should not cause an audible issue.


    One should realize that when measuring the mic preamp, I am also measuring the Lynx Hilo AD and DA converters as well, using the balanced XLR connections. Measuring just the Hilo’s analog loopback using balanced connections puts the THD distortion below 0.0005 percent. But since I am using the headphone output, let’s see the Hilo’s headphone amp measurements in an analog loopback test.


    Lynx Hilo AD DA Converter


    The Hilo still measures at the top of a long list of pro interfaces for AD/DA linearity and transparency. Here I am measuring the “analog” loopback of the converter through the headphone amp. Meaning I have taken the headphone amplifier unbalanced analog output and fed it back into the Hilo’s analog line input. Again using REW as the measurement software, we see a straight pass through with minimal distortion.





    Frequency response.




    Phase response.






    Distortion. Highest THD was at 30 Hz at 0.0043 percent as per cursor. I could get it to go a bit lower if I moved the connection cable around a bit, so I suspect that this is more related to low frequency noise with an unbalanced connection than distortion. Nonetheless, below our audibility thresholds.


    Conclusion – the measurement and recording chain verified with a flat frequency and phase response and low distortion. I won’t be including these measurements in future articles, other than sharing a link to these verification measurements.






    image20.jpegI wrote this book to provide the audio enthusiast with an easy-to-follow step-by-step guide for designing a custom digital filter that corrects the frequency and timing response of your loudspeakers in your listening environment, so that the music arriving at your ears matches as closely as possible to the content on the recording. Accurate Sound Reproduction using DSP. Click on Look Inside to review the table of contents and read the first few chapters for free.












    image21.jpegMitch “Mitchco” Barnett.

    I love music and audio. I grew up with music around me, as my mom was a piano player (swing) and my dad was an audiophile (jazz). My hobby is building speakers, amps, preamps, etc., and I still DIY today.

    I mixed live sound for a variety of bands, which led to an opportunity to work full-time in a 24-track recording studio. Over 10 years, I recorded, mixed, and sometimes produced over 30 albums, plus numerous audio for video post productions in several recording studios in Western Canada.

    User Feedback

    Recommended Comments

    Very nice - thanks!! I have one minor suggestion though.  It is true that "... the 1.5 kHz to 4 kHz range, [is] also the range our hearing is the most sensitive" for totally normal ears.  But a discrete drop in threshold sensitivity centered on 4kHz (i.e. a notch) is the hallmark of noise-induced hearing loss, and a loss of threshold sensitivity as a sloping low pass function affecting frequencies from about 2kHz up typifies presbycusis (hearing loss commonly associated with aging rather than trauma, toxicity, or otologic disease).  So most humans will lose threshold sensitivity between 2 and 4 kHz from some combination of noise exposure and aging as we pass 50+.

    Share this comment

    Link to comment
    Share on other sites

    Create an account or sign in to comment

    You need to be a member in order to leave a comment

    Create an account

    Sign up for a new account in our community. It's easy!

    Register a new account

    Sign in

    Already have an account? Sign in here.

    Sign In Now

  • Create New...