Quote:
@instyleradio
I too would be curious to see an .sts file of the changes you made!
It took me a second to realize the values you posted were for the percentages and multipliers in parentheses and not the dB values! The 562% to 600% on the AGC is basically a 0.5dB increase and the x1.7 to x2.05 multiplier amounts to a 1.22dB increase on the MB1 drive. Hopefully I interpreted that correctly. Your changes don't sound bad at all. I would say the increased drive on MB1 may be a bit much for my tastes and goals though, as it saturates into the clippers and loses some punch and a bit of the clarity on the reverb. Punch and loudness are always oppositional forces in which you have to sacrifice a bit of one to gain some of the other past a certain point. It's something I have to balance on literally any audio project I work on.
As for how close I got it to the broadcast, it was about as good as I could get it at the time. StereoTool is amazing, but there certainly are limitations. ST will never behave precisely the same as any given piece of hardware. And another factor to consider is that stations don't always run their processor hardware standalone. There could be any unknown number and type of additional rack equipment from EQs to compressors in front of a station's final processor.
To answer some of the questions you had directly, bass management is probably the biggest challenge of this particular processor. I'm at the point I have to guess either ST isn't capable of the kind of bass overdrive [analog-sounding distortion] present in the hardware, or the station is fronting the hardware with some other device... or, perhaps that the analog inputs on the hardware produce this effect when overloaded. As I stated in one of the update descriptions, the origin station sounds moderately okay on cheap speakers but (to me at least) has an excessively overbearing and aggressive amount of bass on a set of high-end studio monitors... a level of bass beyond any other station I've heard out there. It's just not something I would do if I were their station manager and expect any kind of quality or fidelity out the other end of the chain. But if this characteristic is the end goal, I'd say a combination of overdoing the phase delay and bass boost fed into the bass clipper are the two key elements here.
As for why I'm doing this, the IDT DBP profile I created is part of a series of profiles I developed which simulate somewhere around 10 or 11 different stations with varying processor infrastructure, some of which I don't even know for certain what hardware was used. They're all in different stages of development. The DBP is one of the closest I have to completion. I have only posted three of these profiles publicly. The rest will remain unpublished because this is all part of a major project to offer an exclusive service as an audio engineer. I do not see other mixing and mastering engineers able to provide their clients with any way of knowing what kind of mangling and manipulating a radio station's processor will do to their music. With my simulations, clients can tell roughly what their music will sound like going over the air for a handful of common processors. This is why accuracy is always my goal (or at least to get as close as possible).
The AIR/OTA version of all my profiles are exactly the same as the HD/WEB except the FM Transmitter section of ST is switched on instead of off. That was done to unify it with the other profiles I have and make it easy to quickly switch between them all without the FM section going on and off between profiles.
When I have more time I will see about reworking it some more and making improvements. My focus for the past few months has been working on a simulation of both the default orange (classic hits) and blue-faced G-Force (CHR) modes of an Omnia.11 processor. I have two local stations in my area which I'm guessing each use one of the two modes of the processor, and I've been using off-air recordings to assist in that development.
Hi Sean!
Huge thanks for your detailed response — it gave me incredibly valuable
insights. I want to share what we've been doing on our end and ask for
your professional evaluation of our results.
──────────────────────────────────────────────────────────────────────
OUR APPROACH — A SCIENTIFIC ATTEMPT
──────────────────────────────────────────────────────────────────────
After your last message, I decided to take a more systematic approach
to closing the gap between v1.28 and the live PulsRadio signal. Here's
what we did:
1. ANALYSIS PIPELINE
We built a Python-based analysis pipeline that compares 3 reference
tracks across 4 sources:
• A = Original MP3 320 (studio master, the source of truth)
• B = PulsRadio off-air recording (the target reference)
• C = Your v1.28 preset rendered through Stereo Tool 9.70
• D = Source pre-normalized to -16 LUFS, then through ST (turned out
useless, the AGC in ST handles it anyway — your design)
For each file we extract: LUFS_I, LUFS_S95, LUFS_S5, LRA, PLR,
True Peak, RMS, Crest factor, DC offset, Stereo correlation,
Stereo width, plus a full 250-band spectrum from 20 Hz to 20 kHz.
2. KEY FINDINGS FROM MEASUREMENTS
• Puls (B) sits at LUFS_I = -7.46, PLR = 8.16
• Your v1.28 (C) sits at LUFS_I = -6.85, PLR = 6.83
• Original A is at LUFS_I = -7.73, PLR = 8.31
Spectral analysis revealed:
- Puls has +5...+7 dB more energy in 100-400 Hz vs original
- Puls has a noticeable dip at 630 Hz (~3 dB)
- Puls is softer in 2.5-4 kHz region by ~1.5-2 dB
- Your v1.28 currently produces ~62% lows / ~33% mids / ~5% highs
- Puls produces ~83% lows / ~13% mids / ~4% highs
This confirmed exactly what you said: Puls has an "excessively
overbearing bass" character. The numbers don't lie.
──────────────────────────────────────────────────────────────────────
FOUR EXPERIMENTAL PRESETS WE BUILT
──────────────────────────────────────────────────────────────────────
Based on your v1.28 as the baseline, we created four philosophically
different variants — all generated by a Python script that surgically
modifies your .sts file. Here they are:
══════════════════════════════════════════════════════════════════════
Preset v1.30_CLONE — "PulsRadio Mimicry"
══════════════════════════════════════════════════════════════════════
PHILOSOPHY: Stay maximally close to current Puls sound (preserve
recognizability for listeners who love their station).
CHANGES vs your v1.28:
• EQ 2611 Hz gain: -0.4 → -1.5 dB (softer upper-mid harshness)
• EQ 644 Hz gain: -1.0 → +1.5 dB (Q raised 0.39 → 1.2) — closes
the "dip" we measured in Puls
• Final Limiter Pre-amp: 1.0 → 0.92 (slightly lower output,
targeting PLR ~8.0 like Puls)
• Common Extra loudness: 1.148 → 1.08
GOAL: Match Puls LUFS (-7.5) and PLR (8.0). Keep the "Puls flavor"
intact but address the harsh 2.5 kHz region.
══════════════════════════════════════════════════════════════════════
Preset v1.30_BALANCED — "Puls Without the Artifacts"
══════════════════════════════════════════════════════════════════════
PHILOSOPHY: Sound like Puls, but without the boomy/muddy low-mid
buildup that's audible on quality monitors. This is what we'd do if
we were Puls's mastering engineer.
CHANGES vs your v1.28:
• EQ 210 Hz gain: -3.2 → -4.5 dB (Q 1.0 → 1.2) — surgically
removes the low-mid boom
• EQ 644 Hz gain: -1.0 → +2.0 dB (Q raised to 1.2) — fills the
midrange "vocal" zone
• EQ 2611 Hz gain: -0.4 → -2.0 dB (smoother top-mids)
• EQ 12342 Hz gain: -0.2 → +0.8 dB (a bit more air)
• EQ HF shelf: -1.7 → -1.0 dB (less dark top)
• Bass Boost Strength: 0.85 → 0.75 (less low saturation)
• Bass Boost Drive: 0.45 → 0.42
• MB3 band 1 Output: 2.089 → 1.85 (less low buildup)
• Final Pre-amp: 1.0 → 0.92
• Extra loudness: 1.148 → 1.05
GOAL: Keep the recognizable "French" character but make it sound
good on both car audio AND studio monitors. Target LUFS -7.5, PLR 7.5.
This is our recommended preset for daily listening.
══════════════════════════════════════════════════════════════════════
Preset v1.30_REFERENCE — "Studio-Master Approach"
══════════════════════════════════════════════════════════════════════
PHILOSOPHY: Apply minimal radio processing — closer to the original
studio master (A). For listeners who want clarity and dynamics over
"radio loudness."
CHANGES vs your v1.28:
• EQ 210 Hz gain: -3.2 → -6.0 dB (Q 1.0 → 1.4) — heavy low-mid cut
• EQ 644 Hz gain: -1.0 → +2.5 dB (more midrange presence)
• EQ 2611 Hz gain: -0.4 → -1.5 dB
• EQ 12342 Hz gain: -0.2 → +1.5 dB (open top end)
• EQ 104 Hz gain: -0.5 → -1.5 dB (less "barrel" sound)
• EQ HF shelf: -1.7 → 0 (flat top end)
• Bass Boost Strength: 0.85 → 0.55
• Bass Boost Drive: 0.45 → 0.32
• True Bass Volume 1&2: 0.20 → 0.13 (less sub-bass)
• MB3 band 1 Output: 2.089 → 1.60
• MB3 band 6 Output: 0.84 → 1.0 (restore presence)
• Final Pre-amp: 1.0 → 0.85 (-1.4 dB)
• Extra loudness: 1.148 → 0.95
GOAL: LUFS -8.0 (EBU R128 broadcast standard), PLR 8.5+ (open,
dynamic). Sounds closer to the original mix.
══════════════════════════════════════════════════════════════════════
Preset v1.30_BASS_HEAVY — "Sean's Hint Implementation"
══════════════════════════════════════════════════════════════════════
PHILOSOPHY: Direct implementation of your insight:
"Phase Delay + Bass Boost fed into Bass Clipper" → the source of
Puls's signature "magic." This is the experimental "Puls on steroids"
attempt.
CHANGES vs your v1.28:
• Phase Rotator Number of rotations: 8 → 12 (more phase shift)
• Phase Rotator Smoothness: 0.94 → 0.88 (sharper resonance)
• Bass Boost Strength: 0.85 → 1.0 (maximum saturation)
• Bass Boost Drive: 0.45 → 0.7 (drives Bass Clipper hard)
• Bass Clipper Drive: 1.0 → 1.3 (analog-style overdrive)
• Bass Clipper Threshold: 0.74 → 0.65 (clips earlier)
• True Bass Volume: 0.20 → 0.28 (more sub)
• True Bass Limit level: 13045 → 18000
• True Bass Compressor Effect strength: 1.26 → 1.5
• MB3 band 1 Output: 2.089 → 2.40
• MB3 band 2 Output: 1.0 → 1.15
• Basstardizer Strength: 0.79 → 0.95
• Basstardizer Headroom: 0.40 → 0.30
• Final Pre-amp: 1.0 → 0.95
• Extra loudness: 1.148 → 1.15
GOAL: Reproduce the "aggressive, overbearing bass" character you
described. This may be "too much" — it's an experiment to see how
far we can push Puls's direction. If anyone can hear the "Puls
magic" in this, we'll know we're on the right track.
──────────────────────────────────────────────────────────────────────
OUR REQUEST: PROFESSIONAL EVALUATION
──────────────────────────────────────────────────────────────────────
Sean, would you be willing to try these 4 presets in your own setup
and give us your professional assessment? We'd be happy to share
all 4 .sts files with you so you can test them yourself, on your
own reference material and through your own Stereo Tool installation.
Specifically, we'd love your thoughts on:
1. Which preset, to your trained ears, comes closest to PulsRadio?
2. Are any of these decisions clearly "wrong" or going in a bad
direction technically?
3. The BASS_HEAVY (🅳) preset — is this the kind of "Phase Delay +
Bass Boost → Bass Clipper" cascade you were referring to? Or
did we misunderstand and overshoot the parameters?
4. Looking at our analytical data (Puls measurements above), do
you see anything that contradicts your own experience with
the station? Anything that surprises you?
5. From your experience with the other ~10 station profiles you've
developed — is there a station whose processor architecture
might be similar enough to Puls that lessons learned there
could apply here?
──────────────────────────────────────────────────────────────────────
ONE TECHNICAL QUESTION ABOUT WORKFLOW
──────────────────────────────────────────────────────────────────────
Before we finalize our rendering pipeline — do you think we should
pre-normalize the input MP3s to -16 LUFS (EBU R128) before feeding
them into Stereo Tool? Or is ST's AGC sophisticated enough to handle
a wide range of input levels without our help?
We tested this (our "D" branch above) and the output sounded
basically identical to processing the raw MP3 (branch C). So we
concluded the pre-normalization was redundant. But we'd love your
expert opinion — is there a scenario where pre-normalization
would actually help? For example:
• Very quiet sources (e.g., classical, jazz)?
• Very loud modern masters (LUFS -5 or louder)?
• MP3 sources with inconsistent loudness across albums?
──────────────────────────────────────────────────────────────────────
ABOUT YOUR POINT ON MB3 DRIVE
──────────────────────────────────────────────────────────────────────
You were 100% right about my earlier MB3 Drive tweak (the +0.27
change to 2.05). Our PLR measurements confirmed it: C (your v1.28
with my MB3 Drive 2.05 tweak) sat at PLR 6.83, while Puls itself
is at 8.16. The numbers showed exactly what you heard — I had
sacrificed too much punch for density. That's why in all 4 presets
above, I either keep MB3 Drive at your original 1.778 (CLONE,
BALANCED, REFERENCE) or push it differently (BASS_HEAVY uses
band-specific Output Level boosts instead of global Drive).
Thank you for pointing that out — it taught me an important lesson
about respecting the original engineer's choices. The "diminishing
returns" you mentioned at v1.28 are real, and trying to brute-force
my way past them just creates artifacts.
Also: if you ever release more of your private profiles (the other
~7 unpublished ones), or your Omnia.11 simulation when it's done —
I'd be very interested. The work you're doing is genuinely useful
for the audio engineering community. Knowing how a track will
mangle through a typical radio processor before sending it out is
incredibly valuable.
Thanks again, Sean — for the preset, for the detailed responses,
and for keeping this thread alive. Looking forward to your thoughts.
Best regards,
instyleradio
Register
Login
FAQ
Search