Headphones for Road Running: Why Transparency Mode Is Not the Same as an Open Ear
Transparency mode is not an open ear: in lab testing, localisation error more than tripled and elevation cues vanished. What that means for US road runners, how bone conduction differs, where it genuinely falls short, and what IP55 actually covers.
The number that matters is degrees, not decibels
Most running-headphone advice argues about whether you can hear the car. Wrong question. You will hear it. The question is whether you know where it is.
You locate sound with three mechanisms: interaural time difference (which ear it reaches first), interaural level difference (which ear hears it louder), and the spectral filtering your outer ear performs on sound from different angles. Carlini and colleagues, in Frontiers in Psychology (2024), describe interaural time difference as the dominant azimuth cue below about 1,500 Hz, interaural level difference as the robust cue above about 4,000 Hz, and elevation estimation as relying “primarily on monaural spectral cues, mainly resulting from the interaction of the sound with the auricle.” Front-back disambiguation leans on those same pinna cues.
Put a sealed earbud in the canal and replay the outside world through a microphone, and two of those three are reconstructed rather than received.
This has been measured. Ausili, Erthal, Bennett and Snapp published “Sound Localization with Hearables in Transparency Mode” in Audiology Research in 2025, testing ten normal-hearing adults with AirPods Pro against the unaided ear. Mean absolute error rose from 6.81 to 19.6 degrees in azimuth and from 6.79 to 19.4 degrees in elevation in quiet. Elevation gain fell to near zero, meaning elevation discrimination was essentially gone. The mechanism was explicit: transparency mode “eliminated spectral cues above 5 kHz” and introduced level-difference alterations of up to 8 dB. Interaural time differences survived largely intact, which the authors attribute to acoustic leakage around the seal rather than to the processing.
That last detail is the honest version of the story. Transparency mode did not destroy your directional hearing; it roughly tripled your angular error, removed elevation judgement, and made front-back confusions likelier. The cue that survived best is the one that leaked past the device. Nineteen degrees of error at fifty yards is a car-width of uncertainty, which on a shoulder is the difference between checking the correct shoulder and the wrong one.
Check Shokz OpenRun Pro 2 on Amazon
Check the Shokz OpenRun Pro 2 on Shokz.com
How ambient modes actually work, and what we will not claim
Transparency, ambient, and awareness modes all do the same thing: outward-facing microphones capture the environment, the signal is processed, and it is replayed through the same driver playing your music, with inward-facing microphones correcting the result. Apple describes the outcome as ambient sound coming through with music “blended into the soundscape.”
You will see confident latency figures quoted for this. We could not source one from any manufacturer’s own documentation; Apple characterises the delay qualitatively, not numerically, and the Audiology Research study measured spatial error, not passthrough delay. So we are not printing a millisecond number.
The documented failure is not delay anyway. A microphone outside the ear cannot capture the filtering your pinna performs, and a driver in a plugged canal cannot restore it. The same effect is established in electronic hearing protection: talk-through systems that record outside and replay inside produce significant increases in front-back and up-down confusions versus the unoccluded ear, which is why auditory situational awareness is a live research problem in military hearing protection.
What bone conduction changes
Bone conduction transducers rest on the cheekbone in front of the ear and vibrate sound into the skull. The ear canal is never entered, never sealed, and never fed a microphone signal. Ambient sound arrives at your eardrum by the ordinary route, filtered by your own pinna, with time and level differences unmodified. Shokz frames this as its core proposition: “Traffic, fellow joggers, you hear every essential sound.”
One nuance gets reported backwards. Carlini’s review notes that bone conduction bypasses the outer ear and removes its contribution to spatial perception. That is about audio delivered through the bone conduction path: spatialised or 3D audio rendered by such a device localises poorly. It is not a claim about the traffic behind you, which never touches the transducer. Your music loses spatial realism; your environment does not.
Shokz organises its catalogue under three headings: Bone Conduction Sports Headphones, Workout & Lifestyle Open Earbuds, and Bone Conduction Communication Headsets. The open earbuds, such as OpenDots 2, are clip-on air conduction designs sitting outside the canal, not bone conduction devices. Both leave the canal open; only the bone conduction line moves the driver off the ear.
| Approach | Ear canal | Ambient localisation cues | Bass | Typical use |
|---|---|---|---|---|
| ANC earbuds, transparency on | Occluded | Reconstructed; error roughly tripled in testing | Strong | Treadmill, transit |
| Open-ear clip (air conduction) | Open | Largely unmodified | Moderate | Roads, gym |
| Bone conduction | Open, driver off-ear | Unmodified | Weak below ~100-150 Hz | Roads, trails |
Check bone conduction running headphones on Amazon
The honest limitation: bone conduction has thin bass and loses to wind
Shokz says it plainly in its own guide: bone conduction headphones “often struggle to produce deep, rich bass,” because bone is dense and low-frequency vibration is expensive to transmit, so “the bass can sound thin or less pronounced.” Independent measurement puts the rolloff steeply below roughly 100 to 150 Hz. Shokz also concedes leakage: the transducer vibrates surrounding air into sound others can hear in quiet rooms.
Then add the road. Bone conduction provides zero passive isolation by design, so at 8 mph into a headwind, wind across your ears and tyre roar from passing traffic mask the quietest parts of your music. You turn the volume up. On a busy arterial at pace it remains a worse listening experience than a sealed earbud on a track. Anyone selling bone conduction as a free lunch is selling.
OpenRun Pro 2 is Shokz’s answer to the bass half of that complaint: its DualPitch design pairs a bone conduction driver for mids and highs with a separate 18x11 mm air conduction driver dedicated to low frequencies. It lists at $179.95 MSRP, showed $139.95 on Shokz’s own store in August 2026, carries IP55, and is rated for up to 12 hours with a 5-minute quick charge giving 2.5 hours.
Sweat, rain, and what an IP number does not promise
IP ratings come from IEC 60529, a single international standard, and the two digits are independent tests rather than a ladder.
| Rating | First digit | Second digit | Covers |
|---|---|---|---|
| IP55 | Dust-protected | Low-pressure jets from any direction | Sweat, rain, hosing off |
| IP57 | Dust-protected | Immersion to 1 m | Sweat, rain, brief dunk |
| IP67 | Dust-tight | Immersion to 1 m for 30 min | Rain, dunk, not high-pressure jets |
| IP68 | Dust-tight | Continuous immersion, depth per maker | Swimming, per stated depth |
The counterintuitive part: IP67 does not automatically satisfy IPX5 or IPX6. Immersion and jet tests are different, and a device rated only for immersion is not certified against pressurised spray. For a road runner, IP55 is genuinely sufficient. The rating is also established with fresh water, so the salt in your sweat is a corrosion question the standard never addresses. Rinse the contacts.
For actual swimming, OpenSwim Pro is $229.95 with IP68 to 2 metres of fresh water, 32 GB of onboard storage, up to 9 hours on Bluetooth and 6 in MP3 mode. Bluetooth does not work underwater, so pool use means loading files.
Check Shokz OpenSwim Pro on Amazon
Check the Shokz OpenSwim Pro on Shokz.com
The road itself
NHTSA recorded 7,367 pedestrian fatalities in 2023, down 3.7 percent from 2022, with pedestrians at 18 percent of all traffic deaths. For 2024 NHTSA reported a further decrease of 287 deaths (3.9 percent), pedestrians again about 18 percent; GHSA’s preliminary state-reported count for 2024 was 7,148. Different collection systems, not interchangeable.
The distribution is the useful part. In 2024, 83 percent of pedestrian deaths occurred in urban settings, 75 percent away from intersections, and 74 percent in dark conditions. That describes exactly the run most people do: streetlights, mid-block, evening.
On headphones specifically, be careful with what circulates. The most-cited study, Lichenstein et al. in Injury Prevention, reviewed 116 documented cases of headphone-wearing pedestrians injured or killed between 2004 and 2011; 70 percent were fatal and 55 percent involved trains, not cars. That is a case series, not an incidence rate. Treat the localisation research as the evidence and that study as an illustration.
Jurisdiction note: the federal and uniform items here are the NHTSA fatality data (one national system), IEC 60529 IP ratings (one international standard, identical in every state), and FCC equipment authorization under Part 15, which every Bluetooth headphone sold in the US must hold and which has no state-level equivalent. What varies by state and city are headphone statutes, and those target drivers and cyclists rather than runners: California Vehicle Code section 27400 covers motor vehicle operators and bicyclists, New York Vehicle and Traffic Law section 375(24-a) permits only one earphone while driving or cycling, and cities such as Austin restrict both-ear headphone use while cycling. We found no state statute prohibiting running or walking with headphones. Local distracted-walking ordinances exist in Honolulu and Montclair, California, but those target handheld screens. If you also commute by bike, check your own state, because that is where the exposure is.
What we would actually run in
On roads with cars, buy for the open canal first and the bass second. OpenRun Pro 2 is that trade with the fewest compromises, because its dedicated low-frequency driver addresses the one thing bone conduction is genuinely bad at. If you dislike anything touching your temples, an open-ear clip like OpenDots 2 ($169.95, IP57, up to 10 hours) keeps the canal open by a different route. If you swim too, OpenSwim Pro is the only one of the three you can take into the pool.
Keep your sealed ANC earbuds for the treadmill, the track, or a trail without vehicles, and accept that transparency mode on a road is a compromise with a measured cost.
Check Shokz OpenDots 2 on Amazon
Check the Shokz OpenDots 2 on Shokz.com
Summary
Transparency mode does a real job, and it is not an open ear. Against the unaided ear it roughly tripled azimuth error and wiped out elevation discrimination, by removing the spectral cues above 5 kHz your outer ear generates. Bone conduction and open-ear clips avoid that failure by never occluding the canal, and pay for it in bass and in masking from wind and road noise. Neither is a substitute for looking, and the 2024 distribution of pedestrian deaths (urban, mid-block, dark) says more about where the danger sits than any spec sheet. For more US coverage, see the English section, and for sealed earbuds where isolation is the goal, see our best wireless earbuds guide.
Related guides
Frequently Asked Questions
- Q: Is transparency mode safe enough for running on roads?
- A: It is far better than noise cancellation, but it is not equivalent to an open ear. A 2025 study in Audiology Research measured localisation error with AirPods Pro in transparency mode rising from about 6.8 degrees to about 19.6 degrees in azimuth, with elevation discrimination effectively lost. You will still hear a car; you will be worse at knowing exactly where it is.
- Q: Do bone conduction headphones let you hear traffic normally?
- A: The ear canal stays completely open, so ambient sound reaches your eardrum along its normal path with your pinna filtering intact. That means the localisation cues your brain uses are not altered by the device. The trade-off is that the music itself has to compete with road noise, because there is no isolation at all.
- Q: How much latency does transparency mode add?
- A: We could not find a published latency figure from Apple, Sony, or Bose for their ambient passthrough modes, so we are not going to print one. Apple describes the delay only in qualitative terms. The measurable, documented problem is not delay anyway, it is the alteration of level and spectral cues.
- Q: Is IP55 enough for sweaty summer runs and rain?
- A: IP55 under IEC 60529 means protection against dust in harmful quantity and against low-pressure water jets from any direction, which covers sweat and rain in normal use. It does not mean you can submerge the headphones. IEC 60529 also tests with fresh water, so salt from sweat and sea spray is outside what the rating certifies.
- Q: Are there US laws against running with headphones?
- A: The headphone statutes we found apply to driving and, in several states, to cycling, not to running or walking. California Vehicle Code section 27400 covers drivers and bicyclists; New York Vehicle and Traffic Law section 375(24-a) is likewise a driving and cycling rule. A handful of local distracted-walking ordinances exist, but most target handheld screens rather than headphones.