This page collects the published Bluetooth codec spectrum findings of one independent measurement author: frequency cutoffs, noise floors and distortion products, measured on a single rig and quoted with sources. The headline rows: LDAC at 303 kbps rolls off around 16.5 kHz while its 606 and 909 kbps settings extend fully to 20 kHz, the macOS AAC encoder cuts off at 15.4 kHz, and aptX dithers its noise floor to around -100 dBFS. These are measured traces, not audibility verdicts; the author''s own caveat travels with every row.
Figures on this page were checked September 2026. Every row cites the measurement author's published article and today's access date.
Sources and methodology
| Element | Detail | Source / as of |
|---|---|---|
| Program | Archimago's Musings, Bluetooth fidelity series, Parts I and II | Published 2023-08-19 and 2023-08-26, accessed 2026-09-22 |
| Receiver | AIYIMA A08 PRO amplifier with Qualcomm QCC5125 Bluetooth chipset | Archimago Part I, 2026-09-22 |
| Sources | Huawei P30 Pro on Android 10; Android 13 Samsung tablet; Windows 11 mini PC; iPhone 11 and 14 Pro; macOS Ventura machine | Archimago Parts I-II, 2026-09-22 |
| Reference and signals | Topping D10s wired DAC as the gold standard; -3 dBFS 1 kHz tone, triple-tone, and 1/10 Decade Multitone 32 at 24-bit/44.1 kHz | Archimago Part I, 2026-09-22 |
One author, one receiver, named sources per link: that is the whole evidence base, and the tables below keep it that way.
Findings by codec
| Codec / mode | Spectral finding | Source / as of |
|---|---|---|
| SBC | Noise floor raised by the codec; "THD+N and TD+N are in the mid -60dB's"; distortion acceptable on simple content, "watch out for degradation with complex signals and higher frequency content" | Archimago Part I, 2026-09-22 |
| aptX | Dithered noise floor "stabilizing... at around -100dBFS", "more noise at higher frequencies", described as alright for a 16-bit codec | Archimago Part I, 2026-09-22 |
| aptX HD (529 kbps) | Lower noise floor than aptX, "a bit more dynamic range by about 10dB", and on the multitone signal "the best thus far with less distortion and noise" | Archimago Part I, 2026-09-22 |
| AAC on Android 10 | "Android AAC encoding has been known to be of poor quality. Indeed, we're seeing this again" on the Huawei P30 Pro | Archimago Part I, 2026-09-22 |
| AAC on iPhone 11 / 14 Pro (256 kbps) | "THD+N scored better than -70dB and TD+N better than -65dB", "similar to the best that LDAC 909kbps achieved in Android"; author's verdict: "better than aptX-HD previously tested and at least equivalent to the higher bitrate LDAC >900kbps on Android" | Archimago Part II, 2026-09-22 |
| LDAC 303 kbps | "The most notable anomaly is the early roll-off with LDAC 303kbps, cut off around 16.5kHz", with elevated high-frequency noise and a "rough patch" beyond 14 kHz before the roll-off | Archimago Part I, 2026-09-22 |
| LDAC 606 kbps | Frequency response extended, multitone components beyond 20 kHz visible in the trace | Archimago Part I, 2026-09-22 |
| LDAC 909 kbps | Extends fully to 20 kHz; a "cluster of distortion around the 650-675Hz tone" also present at 606 kbps | Archimago Part I, 2026-09-22 |
The aptX HD and iPhone AAC rows carry the author's comparative verdicts verbatim; they are his conclusions from his traces, quoted as data instead of restated as ours.
Frequency extension and roll-off
| Codec / mode | Upper frequency behavior | Source / as of |
|---|---|---|
| LDAC 606 / 909 kbps | "Extend fully to 20kHz like the Topping DAC" | Archimago Part I, 2026-09-22 |
| LDAC 303 kbps | Cut off around 16.5 kHz | Archimago Part I, 2026-09-22 |
| SBC / AAC / aptX | Named as the roll-off class the LDAC high modes beat, "without the same amount of SBC/AAC/aptX roll-off" | Archimago Part I, 2026-09-22 |
| macOS Ventura AAC encoder | "Cuts off at 15.4kHz", "even more restrictive than LDAC 303kbps" | Archimago Part II, 2026-09-22 |
What spectrograms show and don't
A spectrogram records what changed in the signal on one rig: where energy stops, where noise sits, where distortion lands. It does not record what anyone heard. The author's own framing travels with his data: "even though the degradation can be measured, how much it impacts audibility will be variable and depends on the listener's hearing abilities," and on the 16.5 kHz cutoff, "might be noticeable by the younger audiophiles". This page archives the traces and leaves the hearing to the reader; per-codec context lives in the LDAC, AAC and aptX HD profiles, and the payload side of the same signals sits in the bitrate bench.
Coverage and gaps
| Codec | Bench status | Source / as of |
|---|---|---|
| aptX Adaptive | "The only one I could not test out", needing a compatible Qualcomm chipset in the source device | Archimago Part I, 2026-09-22 |
| aptX Low Latency, aptX Lossless | Not part of the program | Compilation scope, 2026-09-22 |
| Program breadth | One author, one receiver, Android, Windows and Apple sources; no second spectral program met this page's bar yet | Compilation scope, 2026-09-22 |
| Windows 11 AAC | Tested in Part II's encoder comparison; this page carries only the cutoff and quality rows anchored above | Archimago Part II, 2026-09-22 |
For the delay and payload dimensions of the same codec family, the latency bench and bitrate bench hold the measured record.
FAQ
The frequency-domain fingerprints of a codec on a test signal: where the frequency response rolls off, how high the noise floor sits, and where distortion products appear. One rig produced the findings on this page: a Bluetooth receiver measured against a wired reference DAC, fed 24-bit/44.1 kHz test tones. It shows what changed in the signal, not what a given listener can hear.
In this rig's cross-codec comparison, the author writes that LDAC's 606 and 909 kbps settings "extend fully to 20kHz like the Topping DAC without the same amount of SBC/AAC/aptX roll-off", and LDAC 303 kbps cuts off around 16.5 kHz. The macOS AAC encoder sits lowest at a 15.4 kHz cutoff.
Encoder implementation, not the codec definition. On an Android 10 phone the author found AAC encoding "of poor quality", while iPhone 11 and 14 Pro sources at 256 kbps measured THD+N better than -70 dB, which he attributes, by his own assumption, to hardware-assisted encoding. Same codec, different encoders, different spectra.
The author does rank them on his rig: LDAC 909/990 kbps first ("the best quality codec among the ones tested"), Android AAC last. Every ranking travels with his caveat: "even though the degradation can be measured, how much it impacts audibility will be variable and depends on the listener's hearing abilities." He also notes the 16.5 kHz cutoff "might be noticeable by the younger audiophiles". The traces are data; the hearing is yours.
blafili B3 Bluetooth Receiver
XLR, RCA, optical and coax outputs with an OLED front panel, built on the QCC5125 + ES9018K2M platform. A wired amp or receiver, put on a wireless source.
- Archimago's Musings, "Part I: Comparison of Bluetooth Fidelity - SBC, aptX, AAC, LDAC (Android 10 source)" (August 19, 2023; AIYIMA A08 PRO with Qualcomm QCC5125 receiver, Huawei P30 Pro Android 10 source, Topping D10s DAC as wired reference, 24/44.1 test signals): SBC "THD+N and TD+N are in the mid -60dB's"; aptX "stabilizing the noise floor at around -100dBFS"; aptX HD "a bit more dynamic range by about 10dB" over aptX and "the best thus far with less distortion and noise" on the 1/10 Decade Multitone 32 at 529kbps; "Android AAC encoding has been known to be of poor quality. Indeed, we're seeing this again"; "The most notable anomaly is the early roll-off with LDAC 303kbps, cut off around 16.5kHz"; "the LDAC 606 and 909kbps settings extend fully to 20kHz like the Topping DAC without the same amount of SBC/AAC/aptX roll-off"; "rough patch with LDAC 303kbps just before it rolls off. This is beyond 14kHz"; LDAC 909 "cluster of distortion around the 650-675Hz tone"; "the only one I could not test out was aptX-Adaptive which is only available if the source device uses a compatible Qualcomm chipset"; "even though the degradation can be measured, how much it impacts audibility will be variable and depends on the listener's hearing abilities". archimago.blogspot.com/2023/08/part-i-comparison-of-bluetooth-fidelity.html, accessed 2026-09-22
- Archimago's Musings, "Part II: Comparison of Bluetooth Fidelity - AAC encoder quality (Android 10 & 13, Windows 11, Apple iPhones & Mac)": iPhone 11 and 14 Pro at AAC 256kbps "THD+N scored better than -70dB and TD+N better than -65dB", "similar to the best that LDAC 909kbps achieved in Android"; "macOS Ventura AAC encoder cuts off at 15.4kHz", "even more restrictive than LDAC 303kbps with cut-off at 16.5kHz on the Android 10"; summary verdict "better than aptX-HD previously tested and at least equivalent to the higher bitrate LDAC >900kbps on Android"; "My assumption is that the iPhone is using hardware-assisted encoding". archimago.blogspot.com/2023/08/part-ii-comparison-of-bluetooth.html, accessed 2026-09-22