TL;DR: What This Document Covers
This deep-dive blog post explores practical audio processing techniques for AM radio transmission, focusing on achieving maximum clarity and loudness within strict bandwidth and modulation constraints. It covers AGC, multi-band compression, pre-emphasis, filtering, mono summing, clipping, and asymmetrical modulation — all tailored for low-power AM hobby stations using tools like StereoTool. Detailed implementation tips are provided for macOS, Windows, and Linux users. Technical accuracy and regulatory compliance within EU standards are emphasized throughout.
[Estimated reading time: ~25 minutes]
Introduction
Preprocessing AM radio audio is especially important in low-power transmitters to maximize both coverage and sound quality. Unlike FM broadcasts, the narrower frequency bandwidth of AM signals and their susceptibility to interference present challenges to audio fidelity. However, with proper preprocessing, it is possible to achieve a balanced and clear sound without overmodulation, while also utilizing the transmitter’s full power effectively. The goal is to produce the strongest yet cleanest modulation possible, so that even a low-power AM station’s signal can be heard from far away and stand out from background noise.
This blog post dives deep into AM audio processing techniques and best practices from the perspective of hobbyist stations.
First of All: Thanks and Praise
Since its founding, Radio Blacksmith Knoll (RBSK) has received significant help from Joakim Weckström (JWE) of Realmix Radio. Joakim has spent countless hours fine-tuning his own station’s audio processing and has shared his experiences and proven settings with RBSK – enhanced with practical advice tailored to station-specific differences. Our transmitters come from different manufacturers, and each TX is a unique individual.
A big thank you to JWE – without his selfless contribution, RBSK wouldn’t sound the way it does today. Even with modest power, the station reaches a wide area across Scandinavia and Northern Europe. It’s important to remember that a carrier peak is not the same as a well-modulated transmission. It’s not about raw power, but how you use – and modulate – it.
So What Is This Article About?
This post uses RBSK’s hobbyist station as a reference example, where all audio processing is done via computer. A software-based mixer combines various audio sources (e.g., music, microphone) and routes them into the StereoTool software for processing.
StereoTool is a versatile software audio processor that can perform all the tasks of a traditional hardware audio processor purely in software. StereoTool has been used by major FM and AM stations worldwide, making it an excellent fit even for small-scale AM transmission hobbyists, thanks to its wide range of adjustment options.
Next, we’ll go through how to properly preprocess AM transmission audio – step by step – and what pitfalls to avoid. We’ll also briefly cover the differences in implementation between Mac, Windows, and Linux environments, and address regulations in Finland and the EU that define modulation depth and bandwidth.
Stages and Techniques of AM Audio Preprocessing
AM broadcasting aims for strong and balanced audio within a narrow bandwidth (typically ~5 kHz per sideband). This is achieved through a multi-stage audio chain, typically consisting of the following components:
- Slow Automatic Gain Control (AGC) – maintains consistent volume levels between different program sources.
- Multiband Compression – reduces dynamic range separately across multiple frequency bands.
- Equalization & Pre-emphasis – fine-tunes bass and treble to suit AM transmission.
- Stereo-to-Mono Preparation – combines stereo into mono without introducing phase issues.
- Limiter/Clipper – removes sharp peaks to prevent modulation from exceeding 100%.
- Bandpass Filtering – limits upper audio bandwidth (e.g., 4.5–5 kHz) to avoid adjacent channel interference.
- Additional Fine-Tuning – includes overshoot compensation and asymmetry control when needed.
We will now walk through each stage in detail and explain why they are important in AM broadcasting.
Automatic Gain Control (AGC)
AGC is typically the first step in the processing chain. It reacts slowly to changes in the program material’s loudness and ensures that no source feeds a signal that is too quiet or too loud into the next stages.
For example, if a quiet voice recording follows a music track, the AGC gradually increases the voice level so it doesn’t drown in background noise. Similarly, if a DJ or other program source produces varying levels (e.g., the “DJ rides the fader” effect), a slow AGC smooths out those differences imperceptibly.
The goal is to keep the processor’s input level optimal: high enough to preserve usable dynamics, but not so high that it overloads the following stages in the chain.
AGC works based on very slow gain changes (“molasses level changes”), so it doesn’t create pumping artifacts in the audio. StereoTool includes an adjustable wideband AGC module, allowing you to set the target level and response speed.
It’s recommended to configure the AGC to correct only large and sustained deviations – small dynamic variations should be handled by the later compressor stages.
Pro tip: Make sure none of the input audio sources (e.g., media players, microphones) feed a fully saturated signal into the processor. AGC cannot fix distortion that has already occurred.
StereoTool’s declipper tool can cautiously recover clipped audio by restoring truncated peaks, but it’s always best to keep the original signal clean.
Multiband Compression and Dynamic Control
Once the basic loudness level is under control, the next step is multiband compression. In multiband (or multi-band) compression, the audio is divided into several frequency bands (e.g., bass, midrange, treble), and each band is compressed independently.
Why multiband? Different frequency ranges affect audibility in different ways: for example, the human ear is sensitive to midrange (speech frequencies), while low bass tones require more power to be heard. Multiband compression allows you to balance the sound spectrum so that each part is clearly audible without allowing a single strong bass hit or sibilant to dominate the modulation.
In AM broadcasting, multiband compression is especially valuable because “loudness is king” – a high average loudness improves the station’s presence amid various types of interference. For instance, electrical device noise and the absence of FM’s capture effect mean that an AM signal has to “fight” more aggressively to be heard. A good multiband compressor can raise a quiet vocal line to stand out against a backing track without letting the kick drum trample everything else.
⚠️ Warning: Overcompression Is a Double-Edged Sword
Although it increases average loudness, it also reduces dynamic range and can cause listener fatigue. One of the most common mistakes among hobbyist stations is compressing too hard: everything sounds loud and “mastered in a studio” at first, but over time, constant maximum intensity without pauses becomes tiring to the ears.
Make sure not to overcompress – in a multiband processor, choose moderate compression ratios and sensible crossover frequencies. One well-tested three-band setup uses the following frequency splits:
- Bass: ~25–250 Hz
- Midrange: 250 Hz–3.5 kHz
- Treble: 3.5 kHz–10 kHz
This way, one band targets the low musical tones, another the speech and vocals, and the third handles higher harmonics – allowing each range to be compressed appropriately without pumping artifacts.
StereoTool allows you to select between 2 and 9 bands, but in practice, too many bands can make the sound unnaturally flat; sometimes less is more in this case too.
Noise Management
In AM broadcasting, it’s important that multiband compression does not amplify background noise during silences or quiet passages. A common mistake is to forget a noise gate or broadband expander in the signal chain. When the music softens, a multiband compressor might otherwise push up tape hiss, vinyl crackle, or 50 Hz hum as aggressively as the actual program signal.
To avoid this, a downward expander (i.e., noise gate) is placed early in the processing chain (after AGC). It reduces signal gain when audio drops below a set threshold – effectively acting as a reverse compressor.
In advanced processors like StereoTool, this can even be applied multiband: each band gets its own expander. For example, only the high-frequency band’s hiss is attenuated during silences, while low-frequency hum or AC buzz doesn’t sneak through.
This kind of complex expander + compressor per band configuration mimics the behavior of high-end analog gear (like the Dorrough DAP 310 or Texar Audio Prism) – and StereoTool can replicate this behavior entirely in software.
💡 Tip:
Start with StereoTool’s built-in “AM transmitter” preset. It typically includes moderate multiband settings tailored for AM use. You can then fine-tune the crossover frequencies and compression levels to match your station’s content type.
Always listen to the result on both high-quality headphones and a typical AM receiver – the balance must work in both environments.
Equalization, Pre-emphasis, and Bandwidth Limiting
The audio bandwidth of AM broadcasting is limited, so shaping the frequency response plays a crucial role. On the high end, we must cut off all frequencies above 5 kHz (or 4.5 kHz) to prevent interference with adjacent channels. On the other hand, we want to keep the audio as bright as possible to avoid the “muddy telephone sound” often associated with AM. On the low end, very low frequencies should be rolled off, as they consume modulation headroom without significantly improving the listening experience on typical radio receivers.
High Frequencies and Pre-emphasis
Many AM radios—especially higher-quality models certified under the AMAX standard—include a de-emphasis circuit in the receiver to reduce high-frequency noise. For this reason, in the 1980s, the NRSC pre-emphasis curve was standardized in the U.S. to brighten transmitted audio by approximately 10 dB between 1 kHz and 10 kHz, with a steep cutoff around 10.2 kHz. In Europe, there was no equivalent official AM pre-emphasis standard—partly because the channel spacing here is narrower (9 kHz spacing vs. 10 kHz in the U.S.). However, in practice, the NRSC curve closely resembles the 50 µs pre-emphasis curve used in European FM broadcasts.
Stereo Tool supports NRSC pre-emphasis in its AM profile with a single toggle, and we recommend using it—especially if you know your audience uses high-quality receivers. Pre-emphasis improves the clarity of AM audio—for example, sibilants and consonants in speech content become more audible through background noise.
However, pre-emphasis combined with compression can easily cause sharp treble peaks to overshoot. Therefore, the processor must include a de-esser or another high-frequency control mechanism. StereoTool includes an Advanced Highs Protection feature that tames excessive treble energy before clipping. (This keeps things like harsh sibilance or vinyl crackle from overwhelming the modulation.)
Good pre-processing enhances treble in a controlled way: it adds clarity, but doesn’t let sharp peaks through.
💡 Note:
In Finland, AM pre-emphasis has not been as widely used as in the U.S. The main reason is the narrower AM bandwidth (usually ~4.5–6 kHz). Excessive pre-emphasis can accentuate noise if receivers don’t have matching de-emphasis. Therefore, many European AM stations opt for a fixed EQ curve: cut low bass, slightly boost the presence range (~2–4 kHz) for speech intelligibility, and sharply cut everything above 5–6 kHz. You can think of this as an “AM EQ curve.”
In Stereo Tool, this can be implemented using a manual EQ or with built-in AM presets that include these filters.
Bass and Low-Cut Filtering
Typical AM radios (small portables, car radios, etc.) roll off sharply below ~100 Hz. Therefore, it’s a waste of modulation power to transmit extremely low frequencies at full strength—they’re not audible, but they can cause additional strain on the modulator and create pumping effects.
A common practice is to cut frequencies below 40–50 Hz. This filters out power-line hum, sub-bass thumps, and any DC offset. In StereoTool, the Highpass filter handles this; one user-reported setup uses a 30 Hz Butterworth filter (6 dB/octave) to prevent infrasonic modulation. This also frees up modulation headroom for audible bass (i.e., frequencies above ~50 Hz).
That said, we don’t want the AM broadcast to sound thin. Overly aggressive bass filtering can make music sound tinny or harsh. This is where special techniques help:
- StereoTool’s Immersive Bass feature generates upper bass harmonics.
This means artificial overtones are added on top of very low notes, which small speakers can reproduce. The listener perceives “bass presence” even if the actual 50 Hz note isn’t heard—our ears pick up the 100 Hz and 150 Hz harmonics, and the brain fills in the missing lower octave.
- Another trick is using “phase rotator” all-pass filters in the bass range.
These make the waveform of voice signals more symmetrical, leading to smoother modulation. This doesn’t change the spectral content but helps reduce asymmetry caused by sharp consonants.
StereoTool includes built-in phase manipulation in various parts of its processing (especially in asymmetry control), so external analog phase rotators are not needed.
Bandwidth and Filtering
In Europe, AM channel spacing is around 9 kHz, so the practical upper limit of the audio bandwidth is about 4.5 kHz to avoid adjacent-channel interference.
In North America, with a 10 kHz raster, audio can theoretically reach ~5 kHz, and many stations transmit even wider if the adjacent channel is unused. In the UK, the permitted audio bandwidth was recently increased from 6.3 kHz to 9 kHz to improve quality due to the low number of AM stations.
The exact approved bandwidth that the station can use is mentioned in their license.
For Finnish hobby stations, a sharp low-pass filter at 5 kHz is highly recommended. StereoTool offers extremely steep filters for this—so you can allow, say, 0–5000 Hz to pass and sharply cut everything above that.
⚠️ Steep filters can cause phase delay and transient ringing (a brief overshoot effect after sharp filtering). Therefore, advanced processors include overshoot compensation: they anticipate the filter-induced spikes and attenuate them before they reach the output.
StereoTool handles this automatically, so you can trust that when you set a 5 kHz low-pass filter, modulation peaks remain under control, and you won’t exceed your allowed bandwidth or modulation depth.
(Old-school engineers, without overshoot compensation, would leave a safety margin in modulation level—but this reduces loudness. Fortunately, modern DSP takes care of it far more precisely.)
Converting Stereo Material to Mono (Without Phase Issues)
AM broadcasts are practically always in mono (with rare exceptions like C-QUAM AM stereo, which is backward-compatible with mono). This means that if your program source is in stereo, the channels must be summed together. A simple L+R summing only works ideally if the material is perfectly phase-aligned between both channels.
A common pitfall is that stereo material often contains phase differences—for example, an instrument might be in opposite phase between the left and right channels. When summed, these can partially cancel each other out—resulting in a thin mono signal or even the near-complete disappearance of some sounds.
To prevent this, intelligent downmixing techniques must be used.
Hans van Zutphen, the developer of StereoTool, encountered this problem early in his web radio project: he was streaming hard trance music as a 56 kbps mono stream, and noticed that plain stereo summing robbed the music of its punch. So he developed a stereo-to-mono converter that preserved the full audio content during the summing process.
In practice, this involves phase error compensation—for example, using all-pass filters to shift certain frequencies in one channel before summing, preventing destructive interference.
StereoTool includes a “Downmix to mono” setting, which you should definitely use instead of feeding a pre-mono signal into the processor. The software performs the summing in an optimized way, ensuring the resulting mono signal is just as full-bodied as the original stereo, with no phase cancellations.
A Few More Mono Considerations
Stereo signals may contain decorrelation effects—stereo widening plugins, ghost echoes between channels, etc. When summed to mono, these can cause unpredictable level changes in certain elements of the mix.
That’s why it’s recommended to reduce exaggerated stereo effects before broadcast. In StereoTool’s Stereo Image settings, you can narrow the stereo field slightly before downmixing to produce a more consistent result.
Listening in mono is a valuable test: anything that sounds weird or disappears in mono should be corrected during the playout stage (e.g., by selecting a different track or adjusting the mix).
🎧 Pro Tip:
From time to time, enable mono monitoring of your station’s output (this can be done in most mixing software or in Stereo Tool’s preview function) and listen to what your station actually sounds like to the average listener. That’s how you’ll catch audio content that causes unexpected phase issues when summed to mono.
Limiter and Clipper – Maximizing Modulation Cleanly
To fully utilize the transmission power of an AM transmitter, the modulation depth should be pushed as close to 100% as possible—without exceeding it. This is handled by a limiter/clipper combination at the end of the processing chain.
The limiter is a fast-acting single-band compressor that catches signal peaks and keeps them in check. The clipper, in contrast, is a hard limiter that brutally chops off any signal exceeding a set threshold. In many modern audio processors, these are combined: first, a gentler limiter, and then a harder clipper at the very end to act as a safety net and catch any remaining overshoots.
Why Not Just Use Compression?
Because unlimited audio, especially dynamic music, contains fast transients that would force you to lower the overall signal level if you want to avoid exceeding 100% modulation. By clipping off these peaks, the average and RMS levels can be raised higher—making the broadcast subjectively louder.
As StereoTool’s developers have noted, with a good clipper, StereoTool can produce a louder yet cleaner output than many expensive hardware processors. Clipping is, therefore, an essential part of loudness optimization.
The Price of Clipping: Harmonic Distortion
The trade-off? Harmonic distortion. When you “cut” the waveform with a clipper, it introduces square-wave-like distortion components—especially at high frequencies. Even if you can’t hear these as distortion within the audio band, they show up in the signal spectrum as high-frequency energy that can exceed the allowed audio bandwidth.
This results in adjacent channel interference, often referred to as “splatter.”
To prevent this, a low-pass filter must be placed after the clipper. For example, if your AM transmission is band-limited to 5 kHz, you must ensure that no clipping artifact (like a 7 kHz harmonic) leaks through.
Overshoot and Filter Compensation
However—as discussed earlier—filters can reintroduce overshoot into the signal (due to phase distortion), meaning small peaks could sneak back in even after clipping.
The solution: overshoot compensation, which StereoTool handles internally.
In practice, this means a double clipping stage:
- The main clipper trims the signal to the final modulation level (taking into account pre-emphasis).
- Then the signal is low-pass filtered.
- And finally, a second clipper or limiter grabs any residual overshoots caused by the filter.
This results in a signal that is perfectly brickwalled at 100% modulation depth, without overshooting.
The benefits?
- No negative modulation cutoff (the carrier never drops to -100%, which would make it vanish).
- Positive modulation peaks are also kept within their intended limit.
Settings in StereoTool
Make sure to enable the “Hard limit output” option in StereoTool when used for AM. This ensures that the output signal never exceeds 0 dBFS, which corresponds to 100% modulation (or slightly below, like -0.1 dB, if you want a safety margin).
Also, if possible, enable True Peak metering—this accounts for inter-sample peaks that filters might introduce.
However, in practice, if you’re using StereoTool’s built-in AM presets, these safeguards are already included by default.
Asymmetric Modulation – Is Over 100% Positive Modulation Worth It?
Traditionally, AM transmissions have used symmetric modulation: both the positive and negative half-cycles of the audio waveform modulate the carrier equally, with a maximum depth of ±100%. However, in the United States, as technology evolved and competition intensified, stations sought an even louder sound. It was discovered that allowing the positive half-cycles to exceed 100% (meaning the carrier amplitude goes over 2x its base level) did not cut off the carrier and did not cause the same kind of interference that exceeding 100% negative modulation did (which momentarily removes the carrier and causes spectrum splatter)【source†technical AM broadcasting references】.
The FCC eventually allowed +125% peak modulation on the positive side, provided that the negative peaks do not exceed 100%. This means that audio peaks can, in theory, create 1.25× the power in the form of “over-carrier” transmission. In practice, this can result in a few extra decibels of perceived loudness, at the cost of some additional distortion in the receiver’s demodulation. But in a noisy AM environment, a slight harmonic distortion is often preferable to the listener not hearing the content at all. That’s why many AM stations in the U.S.—especially talk stations, where the human voice is naturally asymmetric—have intentionally used asymmetric modulation.
What About in Europe and Finland?
In Europe and especially Finland, asymmetric modulation has not been traditionally used, and licensing conditions often have not been updated to permit >100% positive modulation. The general rule is that modulation must not cause overdrive—which is usually interpreted as a strict ±100% limit.
Many European transmitters are designed with this symmetric assumption, and they may not be capable of producing more than 100% positive modulation without modification. That said, some hobbyist transmitters (e.g., from PCS Electronics) do advertise the capability for >125% positive peaks—so technically it is possible, even if not officially approved.
StereoTool and Asymmetry
The StereoTool software directly supports asymmetric limiting. You can set an asymmetry percentage up to 200%, though that’s mostly theoretical (virtually no one goes that high). The software ensures that negative peaks are strictly limited to 100%, while positive peaks can be extended to your chosen value.
This is done by injecting a DC offset into the audio stream and using all-pass filters to shift the phase such that one half-cycle gets clipped earlier than the other.
- Benefit: A few decibels more loudness.
- Drawback: Slightly more distortion (over 100% modulation causes non-linearity in diode detectors) and a more demanding modulation baseline adjustment. Also, if something goes wrong and negative modulation exceeds 100%, the resulting artifacts will be very audible—clicks, splatter, and neighbor-channel interference.
Recommendations for Hobbyists in Finland
Unless you explicitly know that asymmetry is permitted—or truly needed—it is safest to stick with symmetric modulation. You can already achieve high loudness through good compression and clipping alone.
However, if you wish to experiment with asymmetric modulation:
- Keep it moderate (e.g., peak at +110%),
- Monitor your modulation monitor closely to ensure negative peaks never exceed 100%,
- Remember that modulation beyond 100% is not linearly beneficial: a +125% peak might only add about 1 dB to your average perceived loudness.
Well-executed processing without asymmetry can sound better than poorly tuned asymmetric modulation. So use it wisely and test carefully.
Practical Implementation on Different Platforms
Hobbyists use a variety of operating systems, so here’s a brief overview of how to set up Stereo Tool on macOS, Windows, and Linux:
macOS
In the case of Radio Blacksmith Knoll (RBSK), everything runs on Mac. StereoTool is available for macOS as a VST/AU plugin as well as a standalone application. One common setup uses Audio Hijack (or similar software), which captures audio from different sources (e.g. Apple Music/iTunes, microphone input), routes them into a virtual mix, and applies the StereoTool plugin in the master output. The processed audio is then directed to the computer’s physical output, which feeds the transmitter.
Another option is using Loopback to create a virtual soundcard: route your sources through Loopback to the standalone version of StereoTool. In the standalone app, select the virtual device as the input, and your real soundcard as the output. A major advantage of macOS is low latency, thanks to the CoreAudio architecture.
Windows
In Windows, a popular approach is using Virtual Audio Cable (VAC). For example, your playback or mixing software can send its output to the VAC input, and StereoTool’s standalone app receives it as input. The processed audio then plays through the line-out of the PC soundcard to the transmitter.
Some users prefer Voicemeeter Banana as a software mixer: it allows for multi-channel routing and comes with built-in virtual audio buses that StereoTool can use. Another method is the Winamp DSP plugin: a somewhat old-school but functional option. You run Winamp in the background with the StereoTool DSP plugin enabled, feeding it from line-in (requires a bit of trickery), and letting the plugin do the processing.
Still, the simplest Windows setup is using the StereoTool standalone version directly. Be sure to choose a low enough buffer size to avoid audio delay. Also: if your CPU load spikes, consider either reducing some of the more resource-intensive settings in StereoTool or using a dedicated machine for audio processing.
Linux
There’s no GUI version of StereoTool for Linux, but a command-line version is available (e.g. a 64-bit ELF binary). It can read and write directly to ALSA or JACK audio devices.
A popular DIY setup is using a Raspberry Pi 4 as the StereoTool processor. Thimeo even provides a prebuilt STAMP image (StereoTool AM Processor) that runs headless. With a HiFiBerry sound card, you get a dedicated AM audio processing box simply by booting from SD card. JACK is particularly useful in Linux: it lets you route streams from internet radio or playlists into StereoTool and out to a sound device, all virtually.
This requires a bit of shell work, but the reward is a very stable, lightweight solution – the Pi can run audio processing 24/7. If you’re not into Linux, an alternative is running the Windows version of StereoTool using WINE. Many users report it works just fine.
Shared Best Practices (All Platforms)
- Ensure your sound card quality is adequate, and input/output levels are correctly calibrated.
- Use a 24-bit signal path whenever possible to retain headroom and resolution throughout.
- StereoTool processes internally in floating point, so internal clipping is rare, but you must avoid exceeding 0 dBFS at the input or output – red lights in your OS mixer are the enemy.
- Avoid unnecessary format conversions. If everything runs on one machine, skip analog detours entirely.
- If you must involve an analog mixer or external hardware at some stage, keep signal levels balanced, cables short, and grounding solid to avoid noise.
Common Mistakes in AM Audio Preprocessing (and How to Avoid Them)
To wrap things up, here’s a list of common pitfalls AM hobbyists often encounter—and how to avoid them:
• Excessive Compression and Clipping
Audio becomes muffled and fatiguing when all dynamics are flattened. Don’t fall into the trap of thinking that “everything maxed out” sounds better. Instead, aim to create the illusion of loudness by tuning attack/release times appropriately and leaving room for the music to breathe.
Remember: too much clipping = distortion + splatter.
Fix: Use the processor’s overshoot control (StereoTool handles much of this automatically), and keep multiband compression ratios reasonable. Monitor your modulation spectrum using a spectrum analyzer if needed—you don’t want peaks beyond 5 kHz or weird sidebands.
• Insufficient Processing (Underprocessing)
The opposite extreme is leaving audio nearly untouched. This might sound okay on studio monitors up close, but in reality, your signal will be quiet and lost in background noise. Listeners may say your station “barely comes through,” even if RF signal strength is fine—because your average modulation level is too low.
Fix: Enable AGC, compression, and limiting as described earlier. Maintain the character of the content, but lift the average loudness significantly. As a rule of thumb:
– In speech, modulation should hover constantly between 70–100%.
– In music, 30–100% depending on dynamics.
Avoid long silent passages unless they’re intentional dramatic pauses.
• Poor EQ Choices
If you forget to boost the treble, AM’s limited bandwidth will make everything sound muddy. But too much boost can lead to harshness and sibilance. Similarly, excessive bass will max out your modulation meter, but listeners will just hear a dull rumble—if anything.
Fix: Thoughtful EQ:
- Cut below 50 Hz.
- Slightly boost 2–4 kHz for vocal clarity.
- Gently rise the high end at ~10 dB/decade up to 5 kHz.
- Use a de-esser or limiter to tame sibilants and harsh transients.
Compare your sound with known commercial AM station recordings by running them through your processor and comparing tonal balance.
• Skipping Stereo-to-Mono Downmixing
Feeding a stereo signal directly into an AM transmitter without ensuring mono compatibility can be disastrous. Some stereo effects may cancel out or vanish entirely.
Fix: Always downmix to mono before the transmitter. Use StereoTool’s downmix-to-mono feature to prevent phase cancellations. And always test your signal in mono.
• Clipping Somewhere in the Signal Chain
This is one of the most common (and nasty-sounding) problems: audio going red-hot at some stage—either digital (0 dBFS+) or analog (overdriven input). Distortion here is irreversible and made worse by further processing.
Fix: Audit your entire audio path, end-to-end:
- Check mixer faders
- Virtual cable levels
- Stereo Tool’s input gain
- Sound card output levels
Make sure nothing is clipping. Leave headroom: Stereo Tool’s Pre Amplifier might need to be set to –3 to –6 dB if your source is already hot.
Pro tip: Listen quietly with headphones—clean audio stays clean at low volumes; clipped audio turns gritty and harsh.
• Modulation Imbalance (especially with Asymmetry)
If you’re using asymmetric modulation, incorrect settings can cause the carrier to cut off (negatives >100%) or push positive peaks into distortion.
Fix: If unsure, set asymmetry to 0% and keep modulation symmetric—it’s the safest route.
If you do use it, adjust StereoTool’s Asymmetry Strength carefully and monitor it with a modulation monitor or oscilloscope. In a trapezoid pattern: the top line can bulge outward, but the bottom must not lift off the zero baseline. Listen closely—excessive asymmetry can cause audible crackling in sibilants.
• Causing Interference to Adjacent Channels
This is both a mistake and a regulatory violation. It happens when your audio bandwidth isn’t properly limited or if your transmitter’s modulator is driven too hard. Symptoms: “splatter” heard on neighboring frequencies.
Fix:
- Set your low-pass filter tight enough:
– In Europe: 4.5–5 kHz
– In the US: 5 kHz (unless adjacent channels are in use)
- Do not transmit stereo pilot tones or other FM artifacts.
- If your transmitter is old or basic, consider adding a physical low-pass filter: NRSC-compliant if you’re on US frequencies, or a slightly tighter one for European bands.
Rule of thumb: Well-processed audio stays in its lane. Poorly processed audio leaks into someone else’s.
Regulations and Standards: What to Consider in Finland / the EU
To close, a few important words about norms and regulations. The technical requirements for AM radio are governed by international agreements and national regulations. In Finland, Traficom (formerly Viestintävirasto) adheres to the Geneva 1975 Medium Wave Plan, which divides the 531–1602 kHz range into 9 kHz channels.
This effectively enforces the previously mentioned 4.5 kHz sideband limit, to avoid adjacent channel interference.
Modulation Levels
Finland doesn’t publish civilian guidelines specifically on modulation asymmetry, so it’s safest to assume that ±100% is the legal limit.
For reference, Ofcom in the UK has previously set a maximum of ±97.5% modulation for AM stations (allowing some margin to avoid overshoot). The US FCC allows +125% on the positive peaks, but this does not apply in the EU unless your license explicitly allows it.
If your station has an official broadcast license for AM transmission, your license conditions likely include:
- Maximum transmission power
- Exact assigned frequency
- Spectral mask or bandwidth limits
For example, your license might require compliance with ETSI EN 302 017, which specifies spectral purity requirements for AM transmitters, including limits on sideband distortion.
Typically, all sideband energy beyond ±9 kHz must be suppressed to at least –50 dB on adjacent channels (value for illustration; consult your actual license for exact figures).
Well-configured audio processing is essential to meet these specs—overshoot limiting and sharp filters help keep your signal within its mask.
License-Free (Low-Power) AM Broadcasting
Many hobby stations operate at very low power without formal licensing, under the concept of short-range devices.
In the EU, SRDs under 30 MHz are generally regulated, but the rules mostly apply to inductive systems—not high-quality AM broadcasting.
In practice, the only truly license-exempt path is to transmit with extremely low power, i.e., <50 mW using a small internal antenna, which limits the range to a few dozen meters.
In the US, FCC Part 15 allows for ~100 mW AM transmitters under specific rules, and many hobbyists take advantage of this. Unfortunately, Finland has no direct equivalent.
Here, any meaningful transmission basically requires a license.
Recommendation:
- Ensure no interference is caused to other radio services
- Choose a frequency not in use within at least several hundred kilometers (AM travels far at night!)
- Set your transmitter’s power to the minimum needed for your target area
- Use proper audio processing to keep your signal within your frequency band
EU-Level Recommendations
The ITU-R BS.412 standard applies to FM deviation limits (not AM), and there is no international loudness standard for AM, because modulation percentage is the defining factor.
For AM, the NRSC-1 (US) and some CEPT recommendations exist. These essentially say what we’ve covered here:
- Control channel width
- Avoid negative overmodulation at all costs
Finally, it’s worth highlighting that proper audio preprocessing isn’t just about sound quality—it’s a vital compliance tool.
When your processor is configured with all the correct limits and filters, your station remains technically “clean” and legally compliant.
You don’t want a regulator (like Traficom) measuring an excessively wide spectrum or overmodulated signal from your transmitter—it could result in warnings or license suspension.
Thankfully, modern DSP audio processors make these risks manageable. For example, StereoTool’s AM profiles include NRSC curves and tight low-pass filters, ensuring your spectral output stays clean.
Summary
AM audio processing is a blend of science and art. On the technical side, it relies on a range of dynamic tools (AGC, compression, limiting) and frequency shaping (filtering, pre-emphasis, EQ) to create a sound that is as loud, clear, and pleasant as possible within the constraints of a narrow bandwidth. At the same time, it must remain within regulatory limits—modulation levels must stay legal, and no energy should spill into neighboring channels.
When done properly, good processing can make even a low-power AM station broadcasting in just 5 kHz sound surprisingly powerful. The bass comes through with surprising punch (sometimes even enhanced by harmonics), the highs are crisp (thanks to pre-emphasis and multiband compression), speech remains intelligible, and music rises above the background noise—all this without introducing noticeable distortion or interference.
As one post on a radio hobbyist forum put it:
“Positive modulation above 100% can increase loudness on the dial at the expense of light audible distortion, but without the effects of splatter.”
This nicely captures the idea that carefully measured boldness in processing can offer real benefits without crossing the line into trouble.
Finally, we strongly encourage experimentation and critical listening. Every content type (e.g., vintage schlager vs. modern pop, talk vs. music) may benefit from different processing profiles.
StereoTool makes this easy by allowing you to save and switch between different presets. For example:
- One preset could be tailored for nighttime long-distance reach (stronger compression, tighter bandwidth)
- Another for local daytime listeners (wider bandwidth, gentler loudness enhancement)
The most important takeaway is to understand the core principles of AM audio processing—and, as outlined in this guide, to avoid the well-known pitfalls that degrade audio quality or violate standards.
With thoughtful preprocessing, even a small AM transmitter can deliver a “big station” sound—clear, powerful, and commanding—without compromising radio tradition or breaking the rules.
We hope this guide helps you fine-tune your station’s sound to its full potential.
Wishing you great transmission days and successful experiments in the world of AM audio processing!
Disclaimer
This article has been written based on extensive research, experimentation, and study of audio processing principles as applied to AM radio broadcasting. While it aims to be technically accurate and helpful, all processing recommendations and interpretations are tailored for very specific use cases involving low-power hobby transmitters. The content reflects the author’s practical experience and understanding and is not to be considered a substitute for official standards, engineering certification, or local broadcasting regulations.
If you spot any factual errors, disagree with any of the presented techniques, or wish to provide corrections or clarifications, you’re warmly invited to contact us via our email.