Assistive Listening Knowledge Base

QuestionPartly answered

How good is the sound each system delivers: bandwidth, codecs and speech understanding?

From https://assistivelisteningknowledgebase.org/questions/sound-quality/ · updated 13 October 2026

  • Affects specifiers, venues, listeners, installers
  • 17 sources
  • Updated
  • First published

Technologies: Auracast, Hearing loops, Infrared, RF/FM, Wi-Fi streaming, Shared

In brief

Loop specifications require a response from 100 Hz to 5 kHz; Auracast uses LC3, with 16 and 24 kHz sampling for Standard Quality and optional 48 kHz modes. FM, infrared and Wi-Fi bandwidth depends on the product and receiving path. Wider audible bandwidth can improve speech understanding for some listeners with hearing loss, but quality ratings and technical specifications alone cannot rank the systems. A manufacturer study favoured Auracast over a loop for sound quality; independent venue research also shows substantial speech benefits from loops.

Why it matters

A system's frequency response, its digital sampling rate and a listener's speech score describe different things. A loop's 100 Hz-5 kHz specification concerns the magnetic signal; LC3's 16, 24 and 48 kHz settings describe sampling. Hearing-device processing can further change what reaches the ear. Studies that ask which sound people prefer also answer a different question from tests that score correctly recognised words or sentences1,2,3,4.

What the evidence shows

The following figures describe specified paths and settings, rather than a controlled comparison of complete systems. Manufacturer figures retain their stated conditions; a specification for an input connector cannot establish the response through a listener's phone, headphones or hearing device5,6,7.

System or pathBandwidth and coding evidenceQualification
Hearing loopETSI describes unmodulated audio-frequency transmission and an IEC 60118-4 target of 100 Hz-5 kHz within ±3 dB of the response at 1 kHz.This is a magnetic-field target. The report allows extension above 5 kHz within its 9 kHz scope, so 5 kHz is not a universal physical cutoff1.
AuracastLC3 encodes separate audio channels; Standard Quality uses 16 or 24 kHz sampling, and High Quality uses 48 kHz.Sampling rate is not the frequency response delivered at the ear. Public-location guidance requires a Standard Quality version of every programme, because High Quality reception is optional2,8.
FM exampleManufacturer Listen Technologies specifies 50 Hz-15 kHz at 72 MHz and 50 Hz-10 kHz at 216 MHz, both ±3 dB, for its LT-800 system.These are the manual's wireless end-to-end table values; its architectural paragraph instead starts at 63 Hz. They do not establish the bandwidth of all FM products5.
Infrared exampleManufacturer Listen Technologies specifies 63 Hz-15 kHz ±3 dB for its LT-82 infrared system.The manual labels these as wireless end-to-end figures. Infrared's light carrier does not set one audio bandwidth for all products6.
Wi-Fi exampleManufacturer Listen Technologies lists 20 Hz-20 kHz ±1 dB at nominal level, 16-bit/48 kHz sampling, and 125 kbps per listening device for its LW-215-04 server.The page does not name the live-audio codec or explain packet overhead. Its network bandwidth figure cannot be treated as an LC3-comparable codec payload rate9.
Wi-Fi input versus listening pathManufacturer Sennheiser lists 40 Hz-20 kHz for MobileConnect's XLR connectors.That figure describes the station's connectors, without establishing the complete phone-to-ear response. The sheet does not name the live streaming codec7.

For Auracast, the Basic Audio Profile lists the following LC3 configurations. Rates are per audio channel: stereo combines separately coded channels. They describe codec payload, before radio overhead and retransmissions10,2.

Sampling rateLC3 bitrate per channelFrame duration and reception support
16 kHz32 kbps10 ms decoding is mandatory for Broadcast Sinks; the profile also lists an optional 7.5 ms configuration10.
24 kHz48 kbps10 ms decoding is mandatory for Broadcast Sinks; the profile also lists an optional 7.5 ms configuration10.
48 kHz80 or 96 kbps; also 124.8 kbps at 7.5 ms or 124 kbps at 10 msThe 80 and 96 kbps configurations exist at both frame durations. These 48 kHz reception configurations are optional10.

Noise and distortion need their measurement conditions. Manufacturer Ampetronic's loop commissioning guidance gives magnetic background noise of -32 dB(A) or better and asks for field and frequency-response measurements across the coverage area. Listen's LT-800 manual gives an A-weighted signal-to-noise ratio of 80 dB with its SQ noise reduction enabled, falling to 60 dB at 72 MHz or 50 dB at 216 MHz with SQ off; SQ requires matching receivers. It specifies total harmonic distortion below 2% at 80% deviation. The infrared LT-82 manual specifies below 2% distortion and A-weighted signal-to-noise ratios above 60 dB in mono or 52 dB in stereo, while warning of noise and distortion near coverage edges. These conditions prevent treating the figures as a single ranking11,5,6.

Digital coding still needs listening tests. The tests for hearing-aid use in Bluetooth SIG's member-funded codec report used error-free transmission and included both speech and music. In its 119-listener online test at 16 kHz, 10 ms LC3 at 32 kbps had similar quality ratings to SBC at 66.4 kbps and better ratings than G.722 at 64 kbps. At 24 kHz, LC3 at 48 kbps outperformed G.722 and SBC at about 96 kbps in tests with 115 online listeners and 20 laboratory listeners. However, a separate 20-expert 16 kHz test rated 7.5 ms LC3 below 10 ms LC3 at the same bitrate, with no significant difference between the 7.5 ms setting and G.722. Hearing status was not reported; these were quality ratings, not speech-recognition trials in listeners with hearing loss. The report allows alternative packet-loss concealment implementations and includes a separate Fraunhofer comparison with 25 listeners under packet loss; the ratings above do not measure those conditions2.

Wider audible bandwidth sometimes improves speech scores. The following studies concern the shared hearing-device path; none compares venue transmission technologies or proves that a particular codec gives better speech understanding12,13,14:

  • EarLens-affiliated research, partly funded by an NIH grant to the company, tested 25 adults with hearing loss and 24 with normal hearing. For the hearing-loss group, extending simulated amplification from 4 to 10 kHz improved sentence reception thresholds by a mean 1.3 dB with spatially separated target and competing speech, and 0.5 dB with diffuse competing speech. The study's authors declare financial interests in EarLens12.
  • A university study funded by the Ontario Research Fund, with hearing aids provided by Signia, tested 24 adults. Maximum audible output frequency averaged 7.5 kHz with extended bandwidth for a 65 dB SPL input. The publisher abstract reports a 4.1% improvement in consonant discrimination over reduced gain above 4.5 kHz, with no significant loudness change; 46% preferred extended bandwidth and 33% had no preference. The full paper was not read13.
  • An Earlens-funded study of 15 adults using its direct-drive hearing devices found a mean 1.1 dB advantage for sentence reception in competing speech with full bandwidth compared with restriction above 5 kHz. Loudness increased by a small but perceptible amount. Eight preferred full bandwidth, six had no preference and one preferred restricted bandwidth, but acclimatisation had used only the full-bandwidth setting14.

Preferences depend on the material and listener. Independent NIH-funded research with 16 children and 16 adults used blinded comparisons of simulated hearing-aid processing. For speech, both bandwidth extended to about 11 kHz and nonlinear frequency compression were preferred over restriction at 5 kHz; the Results section found no significant preference between extended bandwidth and frequency compression. For music there was no overall preference, and greater high-frequency hearing loss was associated with less preference for extended over restricted bandwidth. Eligibility required improved audibility with both extended bandwidth and frequency compression in at least one ear. This is evidence about listening preference, without a venue or installer purchasing decision15.

A manufacturer comparison favoured Auracast. GN Hearing tested its ReSound Vivia hearing aids with 18 listeners with mild-to-moderate hearing loss and 18 with normal hearing in one meeting room. Hearing-aid microphones were off, and participants did not know which path was active. Listeners with hearing loss preferred Auracast in 76% of comparisons, rating quality 7.9/10 versus 5.9/10 for the loop. In a separate manikin measurement with 15 dB flat insertion gain, the telecoil path of the same hearing aid unit rolled off below about 300 Hz and was more than 10 dB below the Auracast path at 200 Hz, consistent with the authors' description of filtering to reduce hum. This compares complete configured paths in one hearing-aid model, rather than isolating LC3; the speech-perception scores were subjective ratings3.

Loops also have measured speech benefits, while music results can differ. Independent research funded by Polish and Norwegian grants tested 22 hearing aid users, with 10 normal-hearing reference listeners, in a meeting room and a reverberant church. Calibrated loops significantly improved speech understanding over listening through hearing-aid microphones to the room sound systems; loop sentence scores reached about 95% for mild losses. This was not a comparison against Auracast. In a separate live orchestral study funded by AGE-WELL Canada, with hearing aids and fittings from Sonova Canada, 20 listeners used study telecoil aids and eight used their own aids without telecoils. In the repeated-measures analysis of 19 telecoil participants, a stage-microphone loop feed did not differ significantly from no feed after correction for multiple comparisons, and the balcony-microphone feed was rated worse. The microphone feed and programme material therefore belong in any sound-quality comparison16,17.

Where sources disagree

The results do not establish a universal technology order. GN's sound-quality comparison favours Auracast, while the independent loop study demonstrates speech benefits against room listening; the comparators and outcomes differ. Bandwidth studies find benefits for some speech tasks, yet music preferences and individual responses vary. These differences limit generalisation from a sample rate or a single listening demonstration3,16,15,13.

The numerical specifications also have different boundaries: a magnetic-field target, a codec setting, a manufacturer's wireless system figure and an XLR connector measurement. The cited Wi-Fi documents leave codec identity unresolved. Bluetooth SIG's codec tests for hearing-aid use used error-free transmission and cannot settle performance under packet loss or establish speech scores for listeners with hearing loss1,2,6,7,9.

What would settle it

Independent comparisons using the same source feed and documented levels, fittings, microphone mixes and receiving devices should measure frequency response, noise and distortion alongside speech scores, listening effort and preference. Several rooms, hearing-loss profiles and implant users would test generalisation beyond the small hearing-aid studies. Codec, bitrate, frame duration and reception errors should be recorded, and the Wi-Fi live-audio format identified. These tests would address the different paths, participants and outcomes left unresolved by the cited evidence3,2,4,12,9.

Planning while it is open

  • Ask which part of the path each bandwidth, noise and distortion figure describes, and record the receiver, settings and codec used. The cited specifications use different measurement boundaries and leave some Wi-Fi coding details unstated5,7,9.
  • Provide Standard Quality for all public Auracast programmes; an optional 48 kHz stream cannot replace the version supported by mandatory receiver configurations8,10.
  • Check the installed loop's field, response and background noise across its listening area, and listen for infrared noise or distortion at coverage edges. See also commissioning tests11,6.
  • Compare the actual speech and music feeds through the receiving paths listeners will use. Collect preference separately from speech-test results, since both feed choice and audible bandwidth affected the cited outcomes17,15,4.

Sources

  1. ETSI TR 103 614 V1.1.1: Electromagnetic compatibility and Radio spectrum Matters (ERM); Hearing Loop System (HLS) intended to assist the hearing impaired in the frequency range 0 Hz to 9 kHz; System Reference Document. ETSI Technical Committee ERM. European Telecommunications Standards Institute, January 2022. Standards body (ETSI, the European telecommunications standards organisation, whose members include manufacturers; a system reference document that asks European spectrum regulators to keep 10 Hz to 9 kHz available for hearing loops, and that supports a radio-equipment standard it says the International Hearing Loop Manufacturers Association is drafting) ↩︎ https://www.etsi.org/deliver/etsi_tr/103600_103699/103614/01.01.01_60/tr_103614v010101p.pdf
  2. Performance characterization of the Low Complexity Communication Codec. Bluetooth SIG Generic Audio Working Group. Bluetooth SIG, 13 June 2023. Standards body (Bluetooth SIG, an industry association funded by its member companies, which owns the Auracast trademark; laboratory tests by SenseLab of FORCE Technology and by Fraunhofer IIS) ↩︎ https://www.bluetooth.com/wp-content/uploads/2023/07/LC3Characterization_WP.pdf
  3. Sound quality comparison of Auracast and teleloop systems in a real-world environment. Jespersen CT, Hashem R, Aranda de Toro MA, Fernández D. GN Hearing (ReSound) white paper M201742-26.02, 17 February 2026. Manufacturer (GN Hearing) ↩︎ https://centerforhearingaccess.org/uploads/resound-jespersen/
  4. ETSI TR 103 225 V1.1.1: Speech and multimedia Transmission Quality (STQ); Transmission quality and speech intelligibility for hearing impaired people. ETSI Technical Committee STQ. European Telecommunications Standards Institute, January 2018. Standards body (ETSI, whose members include telecommunications manufacturers) ↩︎ https://www.etsi.org/deliver/etsi_tr/103200_103299/103225/01.01.01_60/tr_103225v010101p.pdf
  5. User Manual: LT-800-072, LT-803-072 and LT-800-216 Stationary RF Transmitters. Listen Technologies, 3 November 2025. Manufacturer (Listen Technologies, which sells the FM transmitters described as well as infrared, hearing loop, Wi-Fi and Auracast assistive listening products) ↩︎ https://www.listentech.com/wp-content/uploads/2021/01/LT-800_803-User-Manual-2025-11-03-vA4-WEB.pdf
  6. LT-82 & LA-140 User Manual: Stationary IR Transmitter and Stationary IR Radiator. Listen Technologies, 26 September 2022. Manufacturer (Listen Technologies, maker of the ListenIR transmitter and radiator described) ↩︎ https://www.listentech.com/wp-content/uploads/2021/10/LT-82-_-LA-140-User-Manual-2022-09-26.pdf
  7. System Specification (V1.2): MobileConnect Assistive Listening. Sennheiser. Sennheiser electronic GmbH & Co. KG, 2023. Manufacturer (Sennheiser, maker of the MobileConnect Wi-Fi system) ↩︎ https://assets.sennheiser.com/download/assets/MobileConnect_System_Specification_v1.2_EN.pdf/ea3cf3b2e7b711f097eaea55b5d2fb86
  8. Auracast simple transmitter best practices guide, version 3. Bluetooth SIG, 17 October 2025. Standards body (Bluetooth SIG, an industry association funded by its member companies, which owns the Auracast trademark) ↩︎ https://www.bluetooth.com/wp-content/uploads/2022/10/Auracast-Transmitter_Recommendations.pdf
  9. ListenWIFI 4 Channel Wi-Fi Audio Server (NEW). Listen Technologies, undated (read 13 October 2026). Manufacturer (Listen Technologies, maker of the ListenWIFI server described) ↩︎ https://www.listentech.com/shop/product/listenwifi-4-channel-wi-fi-audio-server-new/
  10. Basic Audio Profile, version 1.0.3. Bluetooth SIG Generic Audio Working Group. Bluetooth SIG, 23 June 2026. Standards body (Bluetooth SIG, an industry association funded by its member companies, which owns the Auracast trademark) ↩︎ https://www.bluetooth.com/wp-content/uploads/Files/Specification/HTML/bap_v1-0-3_1782503972/BAP_v1.0.3/out/en/index-en.html
  11. Commissioning an Induction Loop System. Ampetronic. Ampetronic (learning zone web page), undated (read 9 October 2026). Manufacturer (Ampetronic, which makes hearing loop drivers and loop test equipment and sells installer training) ↩︎ https://www.ampetronic.com/info-centre/learning-zone/testing-and-maintaining/commissioning/
  12. Extended high-frequency bandwidth improves reception of speech in spatially separated masking speech. Levy SC, Freed DJ, Nilsson M, Moore BCJ, Puria S. Ear and Hearing 36(5):e214-e224, September 2015. Manufacturer (EarLens-affiliated research, partly supported by NIH/NIDCD SBIR grant R44 DC008499 to EarLens Corporation; the authors declare financial interests in EarLens) ↩︎ https://pmc.ncbi.nlm.nih.gov/articles/PMC4549240/
  13. Speech recognition, loudness, and preference with extended bandwidth hearing aids for adult hearing aid users. Van Eeckhoutte M, Folkeard P, Glista D, Scollie S. International Journal of Audiology (Taylor & Francis), 20 April 2020. Independent (Western University; Ontario Research Fund RE08-072; Signia provided hearing aids; the university receives licensing revenue for the DSL fitting software, returned to research) ↩︎ https://www.tandfonline.com/doi/abs/10.1080/14992027.2020.1750718
  14. Detection, Speech Recognition, Loudness, and Preference Outcomes With a Direct Drive Hearing Aid: Effects of Bandwidth. Folkeard P, Van Eeckhoutte M, Levy S, Dundas D, Abbasalipour P, Glista D, Agrawal S, Scollie S. Trends in Hearing 25:2331216521999139, 19 April 2021. Manufacturer (Earlens funded the direct study costs at the National Centre for Audiology; two authors were Earlens employees and one served on its audiology advisory board) ↩︎ https://pmc.ncbi.nlm.nih.gov/articles/PMC8060758/
  15. Paired comparisons of nonlinear frequency compression, extended bandwidth, and restricted bandwidth hearing-aid processing for children and adults with hearing loss. Brennan MA, McCreery R, Kopun J, Hoover B, Alexander J, Lewis D, Stelmachowicz PG. Journal of the American Academy of Audiology 25(10):983-998, November 2014. Independent (Boys Town National Research Hospital and Purdue University; NIH grants R01 DC04300, P30 DC-4662, T32 DC00013, F32 DC012709 and R03 DC012635) ↩︎ https://pmc.ncbi.nlm.nih.gov/articles/PMC4269381/
  16. Speech Intelligibility in Rooms with and without an Induction Loop for Hearing Aid Users. Kociński J, Ozimek E. Archives of Acoustics 40(1):51-58, 27 January 2015. Independent (Adam Mickiewicz University in Poznań; supported by a grant from Poland's National Science Centre and by Norway through the Norwegian Financial Mechanism) ↩︎ https://doi.org/10.1515/aoa-2015-0007
  17. Evaluating a hearing loop implementation for live orchestral music. McWeeny S, Trainor LJ, Armstrong S, Bosnyak D, Tawfik H, Bruce IC. Ear and Hearing 46(3):808-816, 27 January 2025. Independent (McMaster University; AGE-WELL Canada grant); in-kind hearing aids and fittings from Sonova Canada ↩︎ https://pubmed.ncbi.nlm.nih.gov/39868727/

Change log

  • Published with 17 sources.
  • Corrected bandwidth preference results and codec-test scope; clarified measurement conditions and study sample sizes after checking originals.
  • Second citation check confirmed specification conditions, codec comparisons and listener-study results.

Cite this question

Equal Access Audio. “How good is the sound each system delivers: bandwidth, codecs and speech understanding?” Assistive Listening Knowledge Base. Updated 13 October 2026. https://assistivelisteningknowledgebase.org/questions/sound-quality/ (accessed the date you read it).

Every finding here cites its source, and who funded each source is labelled. This site does not recommend products. Found an error or a missing study? Suggest a correction, or see how suggestions are handled.