First crack is an observable physical event

During roasting, the cracking sound of the beans is usually used by the roaster as a marker for tracking progress. But if treated only as a sound to be heard by ear, part of the information can be lost. The cracking event itself relates to structural change inside the bean as heat, pressure, and accumulated gas change.

Wilson recorded and analyzed coffee cracking sounds, showing the signals of first crack and second crack have different acoustic characteristics. Subsequent research went on to show acoustic signals can be used to track roast state under specific test conditions.

Why sound can become data

Sound is a time-based signal. Instead of recording "a crack was heard," a system can record the start time, duration, number of events, and the signal's spectral characteristics. When this information is recorded alongside temperature and roasting time, a subjective marker becomes comparable data.

Wilson showed first crack has different amplitude and frequency characteristics than second crack, and the rate at which cracks appear also differs. This opens up the ability to use multiple characteristics instead of relying on a single isolated pop.

Not every crack represents the whole batch

A batch contains many beans, and each bean doesn't necessarily change at the same moment. You might hear a few cracks before a clear sequence of cracking forms. So "the first sound heard" and "the crack marker for the whole batch" are two concepts worth distinguishing.

Research by Siebald and colleagues used recordings under various roasting conditions and found acoustic characteristics could relate to bean temperature across most of the process studied. But that same experimental design also showed the signal depends on recording conditions and the roaster.

Cross-referencing with temperature and rate of temperature rise

If only the crack timing is recorded, the data still lacks context. An acoustic marker should be placed alongside bean temperature, ambient temperature, roasting time, and rate of temperature rise. That way, you can know where the event falls on the curve and whether two batches are genuinely similar or just coincidentally had the same timing.

Münchow and colleagues showed roasting conditions relate to sensory results and need to be considered as a system rather than relying on a single isolated marker. So first crack should be treated as an event variable within the roast profile.

Don't tie first crack to a fixed temperature

A common mistake is treating first crack as always appearing at the same temperature. The recorded temperature also depends on sensor position, measurement method, bean type, roasting conditions, and equipment. A value taken from one machine can't automatically become a common standard for every system.

What's more valuable is recording the relationship between crack timing, measured temperature, and rate of temperature rise within the specific system being controlled. Once enough data exists across many batches, a business can then determine a suitable reference range.

Consistent recording requires standardizing how you listen and measure

If using a microphone, placement, distance, recording direction, background noise level, and signal processing need to be kept reasonably stable. If using human hearing, observers need to agree on what counts as the starting marker versus what's just an isolated crack.

Siebald and colleagues used multiple microphones and repeated procedures under controlled conditions to compare signals. This is an important point when moving the idea from research into production control: the measuring equipment is also part of the measurement.

The value of the crack marker lies in the profile, not a single sound

When recorded alongside time, temperature, and rate of temperature rise, first crack can help identify differences between batches. It can also support automation if the signal is captured and processed reliably. But it shouldn't replace the entire body of roasting data.

A good roast profile, then, can treat first crack as a defined event, with a timing, measurement conditions, and a relationship to other variables. At that point, a signal usually heard by ear becomes traceable, analyzable data.

Without an automated system, the roaster can still record the moment a clear sequence of cracks is heard, along with temperature and time. This manual recording doesn't replace acoustic analysis, but it creates structured data for comparing batches later.

A microphone placed near the fan or roasting drum can pick up a lot of background noise. If microphone placement changes between batches, differences in the signal could come from the recording equipment rather than the beans. So a valuable acoustic profile needs to record sensor position, environmental conditions, and signal-processing method.

Acoustic data needs recording conditions

The crack marker should also be recorded as a range if the signal isn't clear. Within a batch, cracking can span a period rather than appear in a single second. Recording the start of the crack sequence, the moment it reaches clear density, and the moment it declines can provide more information than a single marker. This is especially useful when comparing batches with different rates of temperature rise.

References

• Wilson (2014). Coffee roasting acoustics. Journal of the Acoustical Society of America, 135, EL265–EL269. https://doi.org/10.1121/1.4874355
• Siebald et al. (2024). Acoustic condition monitoring of coffee beans during the roasting process. LWT, 199, 116119. https://doi.org/10.1016/j.lwt.2024.116119
• Münchow et al. (2020). Roasting Conditions and Coffee Flavor: A Multi-Study Empirical Investigation. Beverages, 6, 29.