1010Music tangerine: Sampling electric piano

Before I get too far ahead of myself, here are a few tips, tricks and observations while sampling electric piano with 1010Music tangerine.

Velocity levels

In my initial review, I called tangerine “Samplerobot in a tiny box.” As long as you are willing to live within tangerine’s limitations, that impression is true. Samplerobot offers far more sampling options and tools than tangerine, however. Take sample rate and depth, for example. tangerine is limited to 48kHz, 24-bit. Samplerobot supports all the standard rates and depths. Now, 48kHz, 24-bit ain’t bad although you may need to down- or up-sample to another rate or depth depending upon your target.

tangerine offers the most useful sampling control parameters: note range, maximum velocity level, number of velocity levels, and interval between samples. When you specify two or more velocity levels, tangerine computes the specific velocity levels to be collected based upon the maximum velocity level. For two levels and a maximum velocity of 100, tangerine will sample at 50 and 100. For three levels, tangerine samples at 33, 66, and 100.

If you’re sampling an instrument like an analog synth with a “continuous” sound from 0 to 100, that’s OK and convenient. However, you would like to have more control over the specific velocities when sampling “discontinuous” instruments like electric piano, e.g.:

       Range      Source sound
    -----------   ----------------
      1 to  75    Soft EP sound
     76 to 104    Medium EP sound
    105 to 115    Soft EP bark
    116 to 127    Loud EP bark

This is a real-world example of the split points in a sample-playback, electric piano voice.

Given tangerine’s scheme for velocity selection, you cannot find a combination of parameters to capture at least one sample from each source level. Further, you would like to capture at the high end of each source range in order maximize input signal strength. More control is needed.

So, in the end, I captured each source range individually. I merged the source samples into a single directory for convenience — “packing” in 1010Music terminology.

Smpl and Inst tags

This leads to the next problem which is assigning file names such that tangerine can sort out root note, note ranges and velocity levels.

Quoting the manual:

To find the Root Note and velocity information, the tangerine looks in the following places, in the following order until it finds what it needs :

  1. SMPL tags of the WAV files
  2. INST tags of the WAV files
  3. The file names for the WAVs

When parsing the file names, it looks for the following format:

[Text name] + [ - or _ ] + [note number] + [ - or _ ] + [Vel1] + [ - or _ ] + [Vel2]

The [note number] will be interpreted as a decimal MIDI Note number. If [Vel1] and [Vel2] are both found, it will use them as the lowest and highest velocities where this WAV should be applied. If only [Vel1] is available, it will use that as the center point for the range of velocities when this WAV should be applied.

So, set the file names appropriately and away we go. Wrong!

What the manual does not tell you is that tangerine writes SMPL and INST chunks into its sample files. When tangerine loads one of its own sample files, it finds the note and velocity information in the SMPL and INST chunks and it ignores the note/velocity information in the file name, i.e., it never gets to step 3 in the prioritized search sequence above.

I worked around this issue by stripping the SMPL and INST chunks from the sample files. Turns out, if you load a tangerine sample file into Audacity, Audacity discards the SMPL and INST chunks. It’s a simple, but tedious matter of loading all of the sample files into Audacity and then exporting them (via export multiple).

Modulation improvements are needed

If you sample the source instrument dry (no effects), you may be disappointed at first listen. The sampled instrument might sound lifeless without a dynamic, evolving sound.

No problem, you say, add modulation. Unfortunately, tangerine comes up short in a few critical ways:

  • The envelope generator cannot modulate the filter.
  • The LFO cannot modulate effect parameters.

Fortunately, the LFO can modulate Level, Pitch, and Pan yielding tremolo, vibrato and auto-pan effects. Even a barely audible tremolo adds life to a dry electric piano sound.

tangerine effects are spartan: delay and reverb. What really hurts is the lack of LFO modulation. In the case of the delay stage, you cannot module Delay Time. Thus, any attempt at a convincing chorus effect is doomed to fail. Kiss phaser good-bye, too.

Although I’m loath to cooking in effects, I sampled some instruments with a touch of room reverb. This creates a dab of ambience as though the instrument was sampled in a small room.

I hope 1010Music continue to improve tangerine, especially its modulation and effect capabilities.

Copyright © 2024 Paul J. Drongowski

First gig: 1010Music tangerine

Thought I would pass along a few quick comments about gigging with the 1010Music tangerine. I got the urge to use tangerine at my church gig last Sunday. Gotta start sometime, somewhere.

First step on Saturday, I played through Sunday’s music and chose the most appropriate voice for each tune. I wasn’t entirely happy with the woodwind voices, unfortunately. So, I quickly sampled three of my favorite woodwind patches from Yamaha MODX. Thanks to all of my recent experience with tangerine and sampling, I had three new voices ready to go in half-an-hour. It pays to know one’s tools thoroughly!

The next step was pulling together all of the pieces and parts: an Arturia Keylab Essential 49 MIDI controller, the tangerine, a Boss FV30L volume pedal, cables and power adapters. With some spares and charts thrown in, I wondered, “Am I really saving any weight?” 🙂 Keylab Essential plus tangerine in less weight than MODX6, but all those accessories add up fast.

Load-in and set-up went well. No issues. I perched tangerine in the upper right corner of the Keylab. The cables thread through the knobs and sliders to the back — not the most tidy arrangement, but it works. The Keylab Essential is a Mk2 and does not have an expression pedal input. Thus, I routed audio into the Boss FV-30L pedal before hitting my amplifier in MONO. Signal strength was very good.

Sound-wise, the gig was successful. I’m glad that I took the half-hour needed to capture three of my go-to voices. The pipe organ sounds seemed a little wimpy and need boosting. Thank goodness tangerine offers gain individually for each preset.

Performance-wise, the main drawback is tangerine’s small screen. I cannot change voices on the fly as easily as the MODX6 touch screen. tangerine requires careful aim. So much care, that you must avoid hitting one of the eight on-screen trigger pads, each of which play a sample. I learned this lesson the hard way, accidentally triggering a note while selecting a preset voice. Ooops, all eyes on the clumsy keyboardist making noise at an inappropriate time during the service. After that, I minimized the volume pedal after every tune and before selecting a different preset.

Overall, I would rate this first gig experience as a success. I spent Monday sampling a half-dozen classic patches from Roland XV-5050. As I said before, once you’ve been working with tangerine, you can really fly with it.

I renamed and reorganized my presets into “categories.” Woodwind preset names begin with “0”, then “2” for strings, “3” for horns/brass, “4” for pads, “5” for organs, and so on. Once the organization settles down and is final, I will try tangerine’s MIDI program change feature. I will then be able to select a preset from the Keylab Essential over MIDI.

Copyright © 2024 Paul J. Drongowski

MODX EP umph!

This short post is for MODX/Montage people looking for a little more electric piano (EP) “umph”.

I haven’t been all that happy with the factory stock suitcase and “Rd” patches. They seem to be lacking guts. So, I turned to the Genos 70s Suitcase Ballad voice for inspiration. Here are the EQ and effect settings:

Genos voice: 70sSuitcaseBallad

    EQ Low    -4/64  = -0.75dB @ 200Hz
    EQ High   +16/64 = +3.00dB @ 2.0kHz

Reverb: Real Medium Hall +  (MODX/Montage: HD HALL)

    1 Reverb Time          1.7s
    3 Initial Delay Time   22.1ms
    4 High Damp Frequency  8.0kHz
    6 High Ratio           0.8
    13 EQ Low Frequency    800Hz
    14 EQ Low Gain         +6dB
    15 EQ High Frequency   1.0kHz
    16 EQ High Gain        +4dB
       Return Level        64

Insert: Chorus 2  (MODX/Montage: SPX CHORUS)

    1 LFO Frequency        0.34Hz
    2 LFO Depth            29
    3 Feedback Level       0
    4 Delay Offset         0.0ms
    6 EQ Low Frequency     500Hz
    7 EQ Low Gain          0.0dB
    8 EQ High Frequency    7.0kHz
    9 EQ High Gain         +2dB
    10 Dry/Wet             D63>W
    11 EQ Mid Frequency    2.0kHz
    12 EQ Mid Gain         +4dB
    13 EQ Mid Width        1.0
    15 Input Mode          Stereo

The 70s Suitcase Ballad voice has the guts and warmth that I’m looking for. Always steal from the best! I plugged these effect and EQ values into the stock Case 73 Soft Performance.

Yamaha’s signal flow introduces equalization at multiple levels. This makes it difficult to suss or judge the overall EQ curve. However, these settings appear to add a significant bump in the 800Hz to 2.0kHz range. The reverb stage, in particular, has an major effect.

The chorus effect gives a nice shimmer after tweaking. It replaces auto-pan in the original factory patch. I adjusted the chorus dry/wet mix to D16>W in order to get a deeper/richer chorus.

Have fun!

Copyright © 2024 Paul J. Drongowski

A smorgasbord of electric pianos

I’m well into the process of sampling Genos/CSP electric pianos using 1010Music tangerine. I’ll have more to say about the process of sampling the EPs in a later post. Today’s blog is laden with enough detail about EP velocity levels to make your eyes roll. Suffice to say, one needs to know where the levels are in the source material before choosing exact velocities for sampling.

There are so many Genos EP voices that I produced a short list of my favorites:

  Genos/PSR/CSP electric pianos

  Voice            MSB/LSB/PC1 Velocity ranges
  ---------------- ----------- --------------------------------------
  Magnetics          104/0/5   1-75  76-104  105-115 116-127
  Electric Piano 1   0/119/5   1-75  76-106  106-127
  Electric Piano 2   0/122/5   1-60  61- 81   82-112 113-127
  SmoothTine         0/119/6   1-70  71-100  101-127
  SuitcaseSoft       104/7/5   1-46  47- 75   76- 99 100-118  119-127
  SuitcaseWarmth     104/14/5  1-46  47- 75   76- 99 100-118  119-127

I like the warm bell-like character of Magnetics. This voice have been around for ages — Tyros 4! Still, oldies can be goodies. Electric Piano 2 is unique to the CSP/CLP instruments and has a nice, unaggressive character.

I decided to sample: Magnetics, Electric Piano 2, and SuitcaseSoft. Those three voices give me a big enough spectrum of tone colors. I determined source voice velocity ranges by sending fixed velocity values to Genos/CSP and listening for velocity steps. I verified the velocity ranges against UVF meta-data when possible.

For comparison’s sake, I took a brief look at a few MODX voices to identify the waveforms in use (and typical velocity ranges). Here’s a mini-dump:

    MODX/Motif electric pianos

    Performance      Wave
    ---------------- ----
    Vintage 74        EP1  3 levels: Soft, Med, Hard
    R&B Soft          EP3  4 levels: Soft1, Soft2, Hard1, Hard2
    Early 70s         EP1  
    Soft Case         EP3  
    Crunchy Comp      EP3
    Vintage Case      EP3
    Hard Vintage      EP1
    Sweetness         EP1
    Case 75 Amp       EP4  5 levels: p, mp, mf, f, ff
    Dyno Chorus Rd    EP2  4 levels: Soft1, Soft2, Hard1, Hard2
    Dyno Straight MW  EP2
    E.Piano 1         EP3

After A/B testing, MODX EP4 is the same multi-sample as the Genos Suitcase, that is, comparing “Case 75 Amp” versus “SuitcaseSoft” with all effects and EQ turned off. Yamaha added the EP4 waveforms with Montage along with the new, detailed “Rd” and “Wr” multi-samples. Yamaha probably captured EP4 during the long gap between Motif XF and Montage, then tossed it into Montage along with the other new electric piano waveforms.

I also like the MODX Performance “Case 73 Soft”. It uses the newer Rhodes 73 waveforms:

    Part 1                  Part 2
    ----------------------  -------------------------
    Rd73 p         1 -  49  Rd KeyNoise p     1 -  84
    Rd73 mp       50 -  85  Rd KeyNoise mf   85 - 116
    Rd73 mf       86 - 108  Rd KeyNoise f   117 - 127
    Rd73 f       109 - 119  Rd KeyOff mf    Keyoff
    Rd73 ff      120 - 127  Rd KeyOff f     Keyoff
    Rd73 KeyOff  Keyoff     EP2 Soft1+      Keyoff
    Rd73 KeyOff  Keyoff     EP2 Hard1+      Keyoff
    EP2 Hard1+   Keyoff     Rd Soft Keyoff  Keyoff

Yamaha paid far more attention to mechanical noises in Montage/MODX. I’m in the process of punching up this Performance. [Yet another small distractions er, project.]

After a number of “lessons learned”, sampling EPs has progressed well. More to come about the actual process later.

For more about 1010Music tangerine, please see:

Check out my article about Yamaha piano voice programming.

Copyright © 2024 Paul J. Drongowski

1010Music tangerine: A rough edge here and there

After sampling Genos woodwind voices with 1010Music tangerine, I moved on to horns, strings and simple pads:

    SeattleStrings mf 104   5  50  Live       1-127
    Strings mf          0 118  49  Live       1-127
    Strings             0 117  50  Live

    VP Soft           104   0  90  Regular    1-127
    DarkLight         104   3  90  Regular    1-127
    MellowPad           0 117  96  Regular    1-127
    ButterStrings     104   2  51  Regular    1-127

    OrchHornsPad      104  11  62  Live       30-127, 1-127
    SoftHorns           0 117  61  Live       1-127
    MellowHorns         0 119  62  Regular    1-127

That’s enough to cover my basic needs (for liturgical music).

I dialed back all of the Genos EQ and effects, leaving a small amount of room reverb (Genos Real Room+ algorithm) for a bit of ambience. Yamaha “Live” voices are stereo, so I sampled them in stereo. [Duh.] The pads sound rather plain without Genos effects, but I’m adding modulation, delay and reverb on the tangerine itself.

I’m leaving quite a lot of head room when sampling Genos — a good thing. The Symplesound instruments have a bit of graininess which I chalk up to the high level of Symplesound’s multi-samples. The grain shows up when playing big, two-fisted chords and I think something is getting clipped somewhere. I haven’t experienced the same graininess with my own multi-samples, thanks to the head room.

Here are the tangerine voice parameter settings for the SeattleStringsMF preset:

    Level      +3.0dB
    Pitch      +0.00
    Filter     -30.0
    Res         50.0%

    Attack      10.0%
    Decay       10.0%
    Sustain     90.0%
    Release     28.0%
    Velocity    70.0%

    Filter MOD  VEL
    Amount      15.0%
    Filter MOD  KEY
    Amount      30.0%

The level adds a little overall boost. The filter settings bring in LPF through velocity and key scaling. As I said, these voices need sweetening. Reverb is added and I’m still working on chorus…

No filter envelope

My enthusiasm for tangerine has not waned. However, tangerine has a few shortcomings.

The biggest shortcoming is that the envelope does not modulate the filter. Yikes, this could be a deal-breaker for some people. Driving the filter is a pretty basic, needed capability and I hope that 1010Music add a solution in a future update.

What do those numbers mean?

You’ll notice that tangerine states many of its parameters as a percentage. Take the envelope parameters above, for example. Sustain and Velocity as a percentage are meaningful (i.e., percentage of full scale), but what the heck do the time-based parameters mean?

Percentage of what? There must be maximum attack, decay and release times. At the very least, 1010Music should state the maximum times and I’ll do the arithmetic. 1010Music state min and max for the LFO rate, for example.

1010Music need to improve the filter documentation, especially the way filter cut-off interacts with modulation. In particular, I’d like to know when to use negative values to tilt control curves the other way (e.g., key scaling or velocity scaling).

Effects

Based on Web comments, the first version of tangerine’s software implemented a more extensive set of delay and reverb parameters. Early adopters complained about the difficulty of dialing in delays and reverbs. 1010Music responded by “simplifying” the effect parameters.

I think they have gone too far. The reverb parameters are decay, pre-delay and damping — all expressed as percentages. Damping as a percentage makes sense. But, decay and pre-delay are time-based and a percentage is kind of meaningless. I often steal (borrow?) effect parameters from elsewhere and know that a 2.7 second reverb time and 20 millisecond initial delay will get me a large hall. tangerine doesn’t let me dial these basic values in directly and min/max time delays aren’t stated in the manual.

And — I confess — I’m at a total loss with the delay effect. [For now.] I haven’t been able to dial in a chorus effect. The maximum delay is stated, but I can’t get the delay short enough. Chorus needs a delay in the 10 to 50 millisecond range.

Hope these comments are helpful. Fingers are crossed for future updates from 1010Music.

My initial comments about 1010Music tangerine are here.

RTFM (update to an update)

Well, reading the manual can be beneficial. 🙂 You never know what you’ll find.

As to configuring a chorus effect, the manual does specify the maximum delay time: 4 seconds. Thus, the delay time needs to be around one percent (1%) in order to get a chorus-like delay time within 10 to 50 milliseconds.

The real bad news WRT chorus, however, is the lack of LFO modulation. The delay time should be LFO modulated in order to get a dynamic pitch shifting effect. I can hear a difference when delay is set in the 10 to 50 millisecond range, but it ain’t a true chorus without LFO mod. I may experiment with two slightly detuned layers and hear what happens.

So, that’s two big misses for tangerine:

  • The envelope cannot modulate the filter cut-off.
  • The LFO cannot modulate delay time.

Bummer! I hope 1010Music adds these modulation options in a future update. These options are quite essential and expected.

I stumbled across an essential detail about the filter parameter. Filter values between 0 and 100% (positive values) enable a high-pass filter (HPF). Negative filter values enable a low-pass filter (LPF). This explains why I had trouble tuning in the filter and modulation by ear alone. Information like this is someetimes a bear to pull out of the tangerine documentation.

The tangerine manual is pretty good by today’s standards. However, it needs better organization and an index. Some information — like parameter names, description, min and max values — should be collected into a single table. Believe it or not, Yamaha is quite good at organizing this sort of detailed information. See their “Data List” and Synthesizer Parameters PDFs.

Copyright © 2024 Paul J. Drongowski

Peeling a juicy tangerine

A few quick comments and lessons learned with the 1010Music tangerine.

Gotta say, the audio quality is spot on. I haven’t heard any glitches, hiccups, or buzzes. Everything sounds clean. Beauty.

Factory content

Everybody loves factory content! Here’s a quick run-down:

  • There are two acoustic pianos: 10Grand (435MBytes) and 1010Grand (2.24GBytes). 10Grand has three velocity layers (40, 80, 120) taken every three notes. 1010Grand is every note with six layers. 1010Music does not identify the source, but I’m guessing K-Sound? It does sound good.
  • Symplesound consultants provided 20 multi-sampled instruments ranging from AP/EP pianos to synths. You’ll find some lighter weight piano options which will fit into the tangerine’s 64MByte internal memory (no streaming from MicroSD). The Rhodes and Wurlie ain’t bad, but are single strike.
  • Soundtrack Loops provided 800+ loops and kits. The loop names identify BPM and key and are arranged into “construction kit” subdirectories. There are ten kits. In addition to cinematics, you’ll find construction kits for contemporary genres like ambient, tribal, and so forth. Pretty much a sampler for Soundtrack Loops content.
  • Soundopolis is a small collection of FX sounds (e.g., alien spaceship). There is a smattering of percussion (e.g., doumbeck). Have fun. Original source unknown.

These same sounds shipped with the 1010Music Blackbox. tangerine presets are provided for everything, so it’s easy to browse content. It would be nice to know if there are any licensing issues if the sounds are used in a production. 1010Music?

Updating and cloning

My tangerine arrived with firmware version 1.0.1. The current rev is 1.1.6. You’ll need to create an account at 1010Music in order to download the latest release and factory content.

Follow the installation directions. You need to update the internal firmware only once. Thereafter, all of your MicroSD cards will need the latest NANOTANG.BIN file and 1010rsrc.bin file. The tangerine does version checking and tells you if something is missing. It would be great if 1010Music explicitly identified what’s needed to update a cloned MicroSD card. I like to use cloned, working copies and keep the original safe.

It was a good idea…

Plans go astray. My first inclination was to build a multi-sampled instrument from a sampling library. I pulled out an old CD of Q Up Arts Symphonic Fields Forever (SFF). This is a great little collection of symphonic instruments and ensembles a la Mellotron. No loops, but that’s OK.

tangerine likes 48kHz 24-bit. SFF is 44.1kHz 16-bit. Tangerine will do 44.1 with a performance penalty. Enter conversion Hell. On top of sample rate conversion, I want to cover five octaves; SFF is sampled over four octaves. Yada-yada and I decided it was all too much work. I could feel the pull of the black hole that sucks away creative energy…

First sampling session

I chose and auditioned Genos instruments, mainly woodwinds and horns. A few candidates stood out:

    Clarinet&Flutes 104   1  72  Regular  1-74, 75-127
    Clarinet&Oboe   104   1  69  Regular  1-74, 75-127, 1-64, 65-127
    DoubleReeds     104   2  69  Regular  1-64, 65-127, 1-85, 86-127
    Flutes&Oboes    104   2  74  Regular  1-64, 65-127
    FluteSection    104   1  74  Regular  1-127
    OrchWoodwind    104   1  71  Regular  1-85, 86-127, 1-74, 75-127

    OrchHornsPad    104  11  62  Live     30-127, 1-127
    SoftHorns         0 117  61  Live     1-127
    MellowHorns       0 119  62  Regular  1-127

Fortunately, I have the UVF meta-data files for these voices. The meta-data specifies the original velocity splits — helpful information when choosing new sampling levels.

I decided to go simple and sample at a single velocity level. I connected the tangerine to Genos and set the Genos Master Volume level to the 4 o’clock position. [This is a position that I previously ascertained to be +0dBFS.] Parameters on the tangerine side were:

    Rec Input:     Left (tip)
    Gain:          +10.0dB
    Filename:      
    Start note:    C2 (36)
    End note:      C7 (96)
    Sample every:  3
    Vel Layer:     1
    Max Vel:       80
    Note Length:   8
    Release Len:   1
    MIDI Chan:     1
    Rec Thresh:    On
    Threshold:     -60.0dB

The “Regular” Genos voices are MONO, so there isn’t any point in stereo sampling. I recommend setting the Gain first, then the record Threshold. -60.0dB was about 4dB above the quiescent noise floor.

Oh, yeah, remember to kill the reverb, effects and EQ on Genos if you want to sample the dry sound.

Press the record button and let tangerine rip. I cancelled two passes because the level meter crossed into the red zone (probable clipping). The parameters reported here seem satisfactory WRT audio quality.

Hook the tangerine to MODX and use MODX as a MIDI controller. On playback, the audio sounds good! However, the samples are one octave too low. MIDI note 36 should have given me two octaves below Middle C (262Hz). The incoming MIDI may be hitting the Genos voice an octave too low. There are a few ways to fix this and they all require trial and error. Ain’t standards wonderful?

Know thy structure

Nothing like real world experience to test your understanding of UI objects and their relationships. 🙂 I now understand tangerine presets much better. A preset (and its directory) is a container for parameters (preset.xml) and multi-samples. The multi-samples are stored in a subdirectory within the preset directory — presumbly what 1010Music calls “packing”.

The tangerine UI drove me mad when I tried to rename objects. At one point, I gave up and renamed everything on a Windows PC. More learning required…

Well, I could try to remap the samples and fiddle around to make everything right. Or, throw the first burned waffle away and try again from scratch. If I take the latter approach — toss the first attempt away — I have a better chance of outlining a basic procedure and moving forward in the future efficiently.

Copyright © 2024 Paul J. Drongowski

Yamaha announce the Clavinova CLP-800 series

Yamaha have announced the much-anticipated CLP-800 series Clavinova digital pianos.

Personally, I was hoping for a significant bump over the CLP-785, the main reference point for this post. The CLP-785 — and now the CLP-885 — are the flagship “spinet” models. Although I enjoy my CSP-170, my dealer offers an attractive trade-in, trade-up plan which I would exercise given the right motivation. After a quick glance through the Owner’s Manual and Data List, my ardor cooled, sad to say.

Yamaha CLP-845 digital piano

The user interface (i.e., front panel and key functions) and choice of keybed (e.g., GrandTouch and GrandTouch-S) remain largely the same. The piano engine features the same capabilities as the CLP-700 series:

Secondary voices and the XG (GS, GM2) sound set remain the same. Please see the CLP-800 series comparison table as lower-end models may elide specific features and voices.

The CFX Grand and Bösendorfer are the featured pianos, and include Binaural Sampling. The CLP-800 series adds two new acoustic piano voices: Chill Out Piano and Lo-Fi Piano. The same Fortepianos are offered: Scarlatti, Mozart, Beethoven, and Chopin Pianos. All other secondary voices are the same in the CLP-800 series. No big upgrades.

The classical and lesson song lists and rhythm list are the same.

The CLP-800 series Owner’s Manual has many revisions for readability. The different organization makes it difficult to make an exact features by feature comparison between the series. I haven’t spotted any gotta-have new features.

Yamaha have made significant changes in the built-in sound system(s):

Component      CLP-885                  CLP-785
-------------  ------------------------  ------------------------
Amplifiers     (45 W + 30 W + 40 W) × 2  (50 W + 50 W + 50 W) × 2

Speakers       (16 cm with diffuser +    (16 cm + 
               8 cm with diffuser +      8 cm + 
               2.5 cm (dome) with        2.5 cm (dome) +
               Bidirectional Horn) × 2,  transducer) × 2,
               Spruce Cone Speaker       Spruce Cone Speaker

You will find similar changes in spec for other models, should you look at the CLP-800 series comparison chart.

Yamaha have given the DSP effect types a modest boost by adding “Effect types used for a specific Voice”. These effect types are:

    VCM EQ 501             Virtual Circuit Modeling (VCM) vintage EQ
    Comp Distortion        Compressor stage followed by distortion
    Vintage Phaser Stereo  VCM analog vintage phaser
    Stereo Overdrive       Stereo overdrive distortion
    Damper Resonance       Simulates grand piano damper resonance
    Hall 5                 REV-X hall reverb

Yamaha arranger and synth enthusiasts will recognize these effect types. They are DSP algorithms that are typically applied to electric piano and other keyboard voices.

Initial Reaction

My initial reaction is “Big-whoop”. [“Disappointment” for those of you who are not familiar with American sarcasm.]

The big functional change appears to be the use of “diffusers” in the built-in sound system. The CLP-800 Web page has a big chart comparing sound system components. The chart made me feel like I was buying a PA system, not a piano. Yamaha need to find a different way to promote this technology — something to make the technology approachable and cuddly to the average customer. Way too techie!

Frankly, I’m left cold. Since I’m happy with my CSP-170, I cannot find a compelling reason to upgrade to the CLP-800 product line (or the CLP-885, in particular). Guess that trade-up option is going to go unexercised…

Why the underwhelming Clavinova Series 800 update?

We know — from Yamaha’s quarterly call with analysts and investors — that the bottom has fallen out of the world-wide market for home digital pianos. China, in particular, is slow. Possibly, Yamaha did not feel the need or want to drive the home digital piano market too hard at this point in time. Perhaps they are saving bigger updates for the future when the market is more favorable?

This doesn’t mean that the CLP-800 series are bad pianos. I liked the Yamaha CLP-785 very much. As far as I’m concerned, personally, CLP-885 doesn’t give me a compelling reason to upgrade.

Technical footnote

The CLP-700 and CLP-800 series generate high resolution velocity and acceleration data with each MIDI note ON message. Two MIDI continuous controller (CC) messages are generated:

    CC#19    Key acceleration
    CC#88    High-resolution velocity prefix

By “prefix”, I suspect Yamaha mean the high order byte of an extended velocity data value. If you intend to use a “Grand Expression” digital piano with a personal computer-based software instrument (e.g., Modartt Pianoteq, Arturia Piano V, etc.), you will need to filter out or map these “extra” MIDI messages. Check the MIDI Implementation Chart for your model and software instrument (VST).

Copyright © 2024 Paul J. Drongowski

Potential Yamaha P-525 gotcha!

A few recent posts on the Piano World Forum brought a potential Yamaha P-525 issue to my attention. I like the upper-end P-series models, notably the now discontinued P-515 and the P-525. Both models embody good value and sound.

The P-525 improves on the P-515 by adding Grand Expression Modeling (GME). The GME piano engine responds to nuanced playing gestures. That’s a win, in my opinion.

Folks using the P-525 as a piano-action controller have noted two new MIDI continuous controller messages:

    CC#19  Key Acceleration
    CC#88  Expand Velocity

These messages are produced by the Grand Expression Modeling engine. Poly Aftertouch, key acceleration and expand velocity — whatever that is — are sent with every note ON.

That behavior may seems like a desirable feature until you try to drive a piano software instrument like Arturia Piano V (as cited in a Piano World Forum post). These continuous controller messages would also get in the way when using the P-525 as a “bottom” keyboard driving a synth as one might do in a pop/rock gig situation. The latter use-case is what the P-series is for, after all.

Harrumph! As far as I know, there isn’t a way to turn off these messages, sending only MIDI Note ON and Note OFF. Yamaha need to fix this. Until they do, I must issue a strong caution about using the Yamaha P-525 as a piano-action controller.

The P-525 Piano Room implements a Grand Expression parameter. In addition to the Dynamic and Static settings, maybe implement OFF?

Buyer beware! I dare say, this may be an issue for anyone purchasing a Grand Expression model with the intention of driving a piano software instrument (VST). Look before you leap — you may need to filter or remap these extra messages yourself!

Copyright © 2024 Paul J. Drongowski

Chord progressions for Yamaha Chord Looper

Hey, hey, more free content for Yamaha arranger keyboards!

The Yamaha Genos and PSR-SX900 keyboards feature Chord Looper — a chord sequencer that cycles through a user-defined chord progression. Yamaha provides a few example Chord Looper Banks, but folks always wish for more.

So, I collected chord progressions from various public sources and created Chord Looper Banks (and Chord Looper Data files). My Chord Looper Banks have the most common chord progressions and some not-so-common progressions. All are in relatively easy keys and await your creativity.

As an added bonus, I included Java source code for the program that translates Extended ChordPro (CHO text files) to Chord Looper Data (CLD) files. If you don’t care about nerdy stuff and just want to play, then stick with the Chord Looper Banks (CLB) and Chord Looper Data files. 🙂 You don’t need to know anything about Java to use the CLBs and CLDs.

Download the Chord Loops (v1) ZIP file

Check out the README.TXT (in the ZIP file) for more information.

Copyright © 2024 Paul J. Drongowski

Yamaha Montage M internals: First look

Thanks to Dmitry Ko on the Keyboard Corner forum, we have the first solid information about Montage M7 internals! Congratulations, Dmitry!

We owe Dmitry a round of applause and gratitude for providing the first — and extensive — information about Montage M internals. He had to take extra steps to find and identify some components (more than I would have been willing to do with a brand new board, that’s for sure).

Dmitry was kind enough to give me a preview and his analysis is spot on.

Here are my own notes:

IC601 SWP70 #1 (Upper left DM PCB)     YMW832-C    FM-X?

  IC602 Winbond W9812G6KH-5  SDRAM 128 Mbit Parallel 200MHz (8M x 16)

IC401 SWP70 #2 (Middle)                YMW832-C

  IC501 Winbond W9812G6KH-5 SDRAM 128 Mbit Parallel 200MHz (8M x 16)
  IC402 Winbond W9825G6KH-6 SDRAM 256 Mbit Parallel 166MHz (16M x 16)
  IC403 Winbond W29N08GVSIAA NAND flash 8 Gbit (1G x 8)
  IC404 Winbond W29N08GVSIAA NAND flash 8 Gbit (1G x 8)
  IC405 Winbond W29N08GVSIAA NAND flash 8 Gbit (1G x 8)  DM PCB rear
  IC406 Winbond W29N08GVSIAA NAND flash 8 Gbit (1G x 8)  DM PCB rear

IC201 SWP70 #3 (Right)                 YMW832-C

  IC301 Winbond W9812G6KH-5 SDRAM 128 Mbit Parallel 200MHz (8M x 16)
  IC202 Winbond W9825G6KH-6 SDRAM 256 Mbit Parallel 166MHz (16M x 16)
 *IC205 Winbond W29N08GVSIAA NAND flash 8 Gbit (1G x 8)
 *IC206 Winbond W29N08GVSIAA NAND flash 8 Gbit (1G x 8)
 *IC207 Winbond W29N08GVSIAA NAND flash 8 Gbit (1G x 8)  DM PCB rear
 *IC208 Winbond W29N08GVSIAA NAND flash 8 Gbit (1G x 8)  DM PCB rear
  XL201 02238 SWP70 clock

IC701 SSP3                       YJ496A0

  XL701  SSP3 clock
  IC706  TI LV08A 2BK ATPZ
  XL801  SSP3 clock

  XLB02 USB hub clock
  ICB04 GL852G  Genesys Logic 4-port hub USB 2.0 (SSP3 hub)
  Q1102 TPC812S ??? USB2_VBUS

  IC702 SDRAM    No mount?

ICA01 Texas Instruments Sitara AM5728BABCXA

  IC001 NANYA 2245 NT5CC128M16JR-EK DDR3 256MByte
  IC002 NANYA 2245 NT5CC128M16JR-EK DDR3 256MByte
  XLE01 CPU clock

ICB03 GL852G USB 2.0 hub controller (DM PCB rear, CPU hub)

  XLB01 USB hub clock

ICC02 eMMC (Yamaha YN240B0)

ICC01 Fast Ethernet PHY

ICA03 THine THC63LVD1O3D LCD controller (LVDS)

IC904 Texas Instruments TPS659037 Power controller

IC101 Texas Instruments PCM1795 32-bit stereo DAC   DM PCB rear
IC111 Texas Instruments PCM1795 32-bit stereo DAC   DM PCB rear

IC121 Texas Instruments PCM1804 24-bit stereo ADC   DM PCB rear

In a few cases (marked with “*”), copyright labels obscure the chip ID information on the IC package.

Here are a few additional observations.

Yamaha gave the Montage M a major league host CPU: Texas Instruments Sitara AM5728BABCXA. The Sitara is multi-core:

  • Dual-core ARM Cortex-A15 MPU (1.5 GHz)
  • Dual TMS320C66x floating-point VLIW DSP (750 MHz)
  • 2 x dual-core ARM Cortex-M4 co-processors (213 MHz)
  • Dual-core PowerVR SGX544 GPU (532 MHz)

This is a major step up from the single core 800MHz ARM in the original Montage. The Sitara is given twice as much primary memory than the original: 512MBytes of DDR3 RAM.

Thus, folks, you’re going to need a fan. Montage M’s digital logic board (DM) has a substantial metal cover, probably to control RFI. The Sitara has a vanilla heat sink. An opening in the metal cover lets heat escape from the heat sink. The fan draws air from beneath the metal cover/heat sink. This design is different than what I expected, i.e., a very small heat sink plus mini-fan cooler a la Raspberry Pi. This Sitara ain’t no RPi!

The two integrated TMS320C66X DSP cores likely synthesize AN-X. 16 voice AN-X polyphony feels just about right for two TMS320 cores. (Reface CS employs an SSP2 DSP and has 8 voice polyphony.) With compute-intensive AN-X going on, you’re gonna need a fan. I don’t buy the cooling pipe solutions proposed by some and it would be very difficult to position the Sitara in contact with the metal chassis. End of story.

The factory waveform NAND flash is Winbond W29N08GVSIAA. That is the highest capacity ONFI compatible NAND flash made by Winbond. The factory waveforms are compressed (10GB when converted to 16-bit linear format) and reside in 4GBytes of physical NAND flash. User waveforms are uncompressed (3.7GB available capacity) and reside in 4GBytes of (separate) physical NAND flash.

In the original Montage, factory and user waveforms are co-resident in 4GBytes of waveform NAND flash. Yamaha added a third SWP70 tone generator/effects IC and, as we know, dedicated one SWP70 to factory waveforms and another SWP70 to user waveforms. The third SWP70 doesn’t have waveform memory and synthesizes FM-X.

A label covers the top of the eMMC device which provides bulk storage (OS, software, presets, etc.) for the Montage M. The original Montage has a 4GByte eMMC device.

As expected, the SSP2 is out of production and is replaced by SSP3. The SSP3 inherits the digital audio routing and rate conversion chores performed by the original’s SSP2. The SSP3 has its own integrated USB interface and associated USB2.0 4-port hub. All are co-located with the external USB connector.

The THine LVDS handles LCD display duties. The Sitara has serious graphics chops with its dual PowerVR SGX544 GPUs. I haven’t worked out the interface to Montage M’s OLED subdisplay as yet. Genos1 communicates with its subdisplay via SPI.

Thanks, again, Dmitry! This more than enough for the next round of Internet speculations. 🙂

Copyright © 2024 Paul J. Drongowski