Audio Gain Staging for Malaysian Recording Chains

Your recording either hisses with noise when you turn it up or clips into digital distortion on the loudest syllables. That is a gain staging failure, and it happens before you ever touch an EQ. Set the clean voltage path through every device in your chain, from mic to software, so the signal stays above the noise floor and below the clipping point. Get it right once and your raw tracks sit at a level that needs less repair later.

Malaysian broadcast specs give you a ready-made target. RTM and BFM 89.9 both expect delivery that peaks at specific levels. Those numbers are not just for radio engineers. They govern your podcast, your voiceover session, and your home studio takes. The whole job is balancing two failures: a signal so low that preamp hiss becomes the loudest thing in the room, or so high that a single plosive smashes past 0 dBFS and hard-clips irreversibly.

Gain staging is the sequence of level adjustments that carry a mic's tiny voltage up to a usable recording level without frying the signal at any single point. In a Malaysian speech recording chain you have three critical junctions: the mic preamp, the audio interface or mixer output, and the software input trim. Each stage has its own ceiling. Exceed one and you cannot fix it by turning down a later knob. The damage is permanent.

Mic Level To Preamp Gain

Your mic outputs a signal measured in millivolts. A dynamic model like the ones used in RTM's field reporting rigs might push 1.5 mV at normal speech distance. A condenser, common in BFM 89.9 studios, can hit 10 mV or more. The preamp multiplies that voltage up to line level without introducing its own noise. Set preamp gain too low and you amplify the preamp's self-noise later in the chain. Set it too high and a sudden laugh or a slammed table clips the input stage itself. The practical sweet spot for speech is a preamp output averaging around -18 dBFS on your interface meter, with peaks no higher than -10 dBFS.

Interface Output And Software Trim

Your interface converts the analogue signal to digital. If the preamp stage is clean, the interface passes it through. Do not add gain here. The software trim, often a fader or input gain knob in your DAW, is the last chance to adjust level before recording. This is where you set the final ceiling. For a speech recording that will later be compressed and limited to a podcast loudness spec, record with peaks at -6 dBFS. For a raw take destined for further production, hold peaks to -12 dBFS. The BFM 89.9 production spec for unprocessed interview audio is -12 dBFS peak. RTM's broadcast delivery target for finished programmes is -9 dBFS maximum true peak, but that is a delivery figure, not a recording target. Record lower and raise it later in the mix.

Headroom is the gap, measured in decibels, between your loudest peak and the point where the system distorts. In digital audio the ceiling is 0 dBFS. There is no forgiving tape compression past that line. A signal that hits +0.1 dBFS clips instantly. The resulting waveform flattening sounds like a sharp crackle on consonants. Malaysian radio engineers call this "pecah," and a track with it will be rejected by any station.

The noise floor is the other boundary. Every electronic component generates a low-level hiss. In a typical untreated home recording space in Malaysia, the ambient noise floor sits around -50 dBFS with an air conditioner running. A mic signal recorded 40 dB below that is unrecoverable. You cannot separate the voice from the hiss later. The trade-off is real: record too hot and a single clipped syllable ruins the take. Record too low and the noise floor rises when you normalise the file. For voice, 12 dB of clearance is the minimum. 18 dB is safer.

Signal flow gain is the discipline of tracing level changes through every physical and digital connection. A common mistake in Malaysian home studios is plugging a mic into a mixer, setting the channel gain, then adjusting the master fader to hit a target in the DAW. That master fader is a second gain stage. If the channel preamp is already distorting, pulling down the master fader does not remove the distortion. It sends a quieter distorted signal to the computer.

The correct signal flow is: mic to preamp, preamp to interface line input, interface to DAW. At each point, confirm the level with a meter. Do not rely on the DAW's waveform display alone. Use the interface's hardware meter or a dedicated metering plugin. Set the preamp so the interface meter shows an average level around -20 dBFS during normal speech. Check the DAW's input meter and adjust only the software input trim to set the final recording level. If you need more level at the DAW, go back to the preamp and increase it there. Never compensate for a weak preamp setting by boosting the software trim. That amplifies the preamp's noise floor alongside the signal.

Organisation	Recording Target	Delivery Peak	Typical Use
BFM 89.9	-12 dBFS	-9 dBFS	Unprocessed interview audio
RTM	-18 dBFS (pre-mix)	-9 dBFS (true peak)	Finished programme delivery
Podcast (LUFS-based)	-6 dBFS	-1 dBFS true peak	Loudness-normalised speech

Podcast gain staging differs from broadcast because the final loudness target is measured in LUFS, not peak level. Most podcast platforms expect an integrated loudness of -16 LUFS for stereo files, with a maximum true peak of -1 dBFS. That does not mean recording at -1 dBFS. It means recording with enough clearance that your compressor and limiter can work without amplifying preamp noise. A raw podcast voice recording should peak at -6 dBFS. That gives you 5 dB of space below the delivery true peak limit.

Malaysian podcasters recording in untreated rooms face a specific challenge. The noise floor from ceiling fans, traffic bleed, and fluorescent light hum can sit as high as -45 dBFS. If you record with peaks at -12 dBFS and later raise the track to -16 LUFS, the noise floor rises to -29 dBFS and becomes audible in headphones. The solution is not to record hotter. Treat the room or use a closer mic placement. A dynamic mic at 10 cm from the mouth, with the preamp gain set for -6 dBFS peaks, yields a signal-to-noise ratio that survives the loudness lift. Gain staging cannot fix a bad room. It can only stop you from making it worse.

Speech headroom requirements are tighter than music because the dynamic range of the human voice is wider and more sudden. A speaker can jump from a murmur to a shout in a single sentence. Setting headroom for the average level guarantees clipping on the peaks. Setting it for the peaks means the quiet parts disappear into the noise floor. The fix is a three-step metering routine.

Step One: Set The Peak Ceiling

Have the speaker deliver the loudest passage they expect to perform. Set the preamp gain so that passage peaks at -6 dBFS on the interface meter.

Step Two: Check The Quiet Floor

Have them speak at their quietest conversational level. Confirm that the meter still reads above -24 dBFS. If it drops below that, the noise floor will be a problem. Move the mic closer. Do not increase the preamp gain.

Step Three: Test And Listen

Record a 30-second test and normalise it to -16 LUFS in your DAW. Listen on headphones for preamp hiss or room tone during the pauses. If you hear it, you need less distance between mouth and mic, not more gain. The BFM 89.9 production spec for speech headroom is -12 dBFS peak, which matches this method when the loudest passage is set to -6 dBFS and normal speech sits 6 dB lower.

RTM's broadcast delivery target is a true peak of -9 dBFS, measured with an ITU-R BS.1770 compliant meter. That is a finished-mix figure, not a recording target. The RTM chain assumes a production path that includes compression, limiting, and level normalisation before the signal reaches the transmitter. If you deliver a raw recording at -9 dBFS peak, the station's processing will push it into clipping. Record at -18 dBFS to -12 dBFS and let the final limiter handle the last 3 to 9 dB.

BFM 89.9, as an independent talk station, runs a different workflow. Their production spec for unprocessed interview audio is -12 dBFS peak, and they normalise to -23 LUFS for broadcast. This sits closer to the EBU R128 spec than to older peak-based norms. For a Malaysian podcaster or independent producer, the BFM 89.9 reference is the more practical one. You can record at -12 dBFS, edit, and deliver without a heavy mastering chain. The station's own processing handles the final loudness adjustment. If you record for podcast distribution, use -16 LUFS integrated loudness as your target and set your recording headroom to -6 dBFS peaks.

Every gain setting is a decision between noise floor and clipping risk. There is no perfect number. A Malaysian voiceover recorded in a soundproofed studio can safely use 18 dB of clearance and still have a noise floor below -60 dBFS. The same voice recorded in a shoplot with a metal shutter and a standing fan needs 6 dB, or it will be buried in ambient noise. The trade-off is that the shoplot recording is one loud consonant away from clipping. Compensate with a pop filter, a shock mount, and a compressor set to catch only the fastest transients.

Your monitoring setup determines whether you can hear the trade-off accurately. Closed-back headphones are mandatory for tracking. Open-back headphones or nearfield monitors are for mixing. If you set gain while listening on laptop speakers, you will not hear the noise floor and you will record too low. If you monitor through a cheap headphone amplifier that adds its own hiss, you will think the noise floor is worse than it is and record too hot. Use a pair of closed-back headphones with a flat frequency response, driven directly from the interface's headphone output. Bypass any software monitoring effects. Listen only to the raw input.