90% of viewers over 60 in a 2,126-person German broadcaster survey said they often or very often have problems understanding speech on TV (Torcoli et al., IBC 2021). In the UK, 70% of an online audience panel already reported problems hearing what was said on TV back in 2010 (Hearing Health Matters 2011). The BBC drama Jamaica Inn collected 546 complaints after its opening episode and 252 more after the second (Press Association, April 24, 2014), while 430 million people worldwide now need rehabilitation for disabling hearing loss (WHO fact sheet, March 3, 2026). We aggregated data from BBC R&D, German broadcaster WDR, the World Health Organization, RNID, the Audio Engineering Society, IEEE, the Journal of Speech, Language, and Hearing Research, and a dozen other primary sources. Subtitle habits are covered in our separate captions and subtitles statistics roundup; this page tracks the audibility problem itself, its measured causes and the fixes that have been tested.
TL;DR
- 68% of WDR survey participants often or very often struggle to understand TV speech, and 90% of those over 60 do (Torcoli et al., IBC 2021).
- 70% of a UK online panel reported problems hearing TV speech, rising to 76% among over-65s without internet (Hearing Health Matters 2011).
- Jamaica Inn drew 546 complaints after its opening episode and 252 after the second (Press Association, April 24, 2014).
- SS-GB fell from 6.1 million to 3.9 million viewers between episodes one and two after mumbling complaints (Jewish News, February 28, 2017).
- Audibility accounts for 60% of the sound complaints the BBC receives (BBC Academy, loudness guide).
- Loudness-based and speech-based normalisation of the same films can differ by up to 14 dB (Lund and Skovenborg, SMPTE 2014).
- Small flat-TV loudspeakers can drop about 30 dB per octave below 200 Hz (Astolfi et al., IEEE Transactions on Broadcasting 2021).
- 73.7% of Casualty accessible-audio respondents said the slider made the content more enjoyable or easier to understand (BBC R&D 2020).
- 430 million people worldwide need rehabilitation for disabling hearing loss (WHO fact sheet, March 3, 2026).
- Over 18 million people in the UK are deaf, have hearing loss or tinnitus (RNID 2026).
- 55% of US adults aged 75 and older have disabling hearing loss (NIDCD 2024).
1. How Many Viewers Struggle: Surveys by Country and Age
The striking thing about the survey record is not the headline share but how steeply it climbs with age. In the WDR online test, 90% of participants above 60 and 98% of those above 80 said they often or very often struggle with TV speech, against 68% of the full sample. That sample skewed older (42% were aged 61 to 80), so the 68% is not a national prevalence figure, but the age gradient matches the UK picture measured a decade earlier.
The UK data comes from a 2010 audibility study that used the BBC’s Pulse online panel plus a paper diary for older people who do not use the internet. The gap between “occasionally” and “always or often” matters: most viewers hit problems now and then, while a smaller core (11% online, 17% among offline over-65s) hits them routinely. That core is the audience most likely to write complaints. The full German survey design is in the Dialog+ field-test paper, and the UK split is summarised by Hearing Health Matters.
| Metric | Value | Source |
|---|---|---|
| WDR online survey participants who often or very often have problems understanding speech on TV (n = 2,126, Germany, 2020): all ages / above 60 / above 80 | 68% / 90% / 98% | Torcoli et al., Dialog+ field tests, IBC 2021 |
| Average daily reporting sample of the BBC Pulse online panel used in the 2010 UK audibility survey | around 8,000 | IPS, VLV Audibility of Speech on Television Project, 2011 |
| UK online panel adults vs UK adults 65+ without internet (n = 508) reporting problems hearing what is said on TV | 70% vs 76% | BBC audibility study 2010, reported by Hearing Health Matters 2011 |
| UK online panel adults with problems “always or often” vs “occasionally” | 11% vs 59% | Hearing Health Matters 2011 |
| UK adults 65+ without internet with problems “always or often” vs “occasionally” | 17% vs 59% | Hearing Health Matters 2011 |
| UK adults who cannot always understand clearly what is said on TV (two 2010 to 2011 studies combined, one co-run by hearing-aid maker Widex with Channel 4) | 71% | Hearing Health Matters 2011 |
| Viewers reporting difficulties understanding on-screen dialogue (vendor-commissioned survey, sample size not published) | 84% | Xperi (DTS) own survey, Clear Audio report |
| UK adults who generally prefer TV subtitles on (YouGov Profiles, 35,000+ Britons) | 30% | YouGov, October 26, 2026 |
Most recent available data for the UK audibility split: BBC audibility study, 2010. We found no larger UK dialogue-difficulty survey published since, which is why the German broadcaster survey is the more recent age breakdown.
2. Complaints to Broadcasters: Jamaica Inn, SS-GB and the BBC Mailbag
Complaint counts are a lagging, noisy signal, yet the two BBC dramas below show what happens when the core audience from section 1 is pushed past its limit. Jamaica Inn drew 546 complaints after the opening episode alone, and the BBC raised dialogue levels for the remaining episodes, but 252 people still complained after episode two (Press Association, April 24, 2014, via WhatToWatch). Over the same stretch the drama lost 1.6 million of its 6.1 million launch viewers.
Three years later SS-GB repeated the pattern on a smaller complaint scale but with a larger ratings drop. The BBC said it would “look at the sound levels on the programme in time for the next episode” (The Telegraph via Yahoo News, February 20, 2017), yet episode two shed more than two million viewers. In 2011 the BBC had already admitted that bad sound was the second most complained-about topic on its website (PSNEurope via SVG, September 25, 2014).
| Metric | Value | Source |
|---|---|---|
| Jamaica Inn complaints after the opening episode (April 2014) | 546 | Press Association via WhatToWatch, April 24, 2014 |
| Jamaica Inn complaints after the second episode, broadcast April 22, 2014, despite raised sound levels | 252 | Press Association via WhatToWatch, April 24, 2014 |
| Jamaica Inn audience at launch vs viewers lost by episode two | 6.1 million vs 1.6 million lost | Press Association via WhatToWatch, April 24, 2014 |
| Jamaica Inn complaints over the full three-night run, by the BBC’s own account | more than 1,000 | PSNEurope via SVG, 2014 |
| SS-GB complaints received by the afternoon after the February 2017 premiere | fewer than 100 | The Telegraph via Yahoo News, February 20, 2017 |
| SS-GB audience, episode one vs episode two | 6.1 million vs 3.9 million | Jewish News, February 28, 2017 |
| Share of BBC sound complaints that concern audibility | 60% | BBC Academy, Why is loudness such an issue for TV viewers? |
| Share of BBC sound complaints caused by differences in loudness range (the gap between quiet and loud passages) | 20% | BBC Academy, Why is loudness such an issue for TV viewers? |
Context note: SS-GB audience figures come from Jewish News. Final Jamaica Inn totals differ between outlets; see the Data watch note at the end of this page.
3. Measured Causes: Background Music, Sound Capture and Cinema-Style Mixing
Press coverage of these disputes tends to blame actors’ diction, while the measurements point mostly at the mix and the recording chain. When the VLV, BBC and RNID project analysed 21 programmes in detail, it concluded that the majority of audibility problems came from the way sound is captured, and that added background music makes audibility even worse (IPS, March 2011). That is a production-chain finding, not a verdict on diction alone.
The laboratory work explains why music is such an effective spoiler for ageing ears. Middle-aged listeners need a measurably better signal to follow speech against music and competing talkers than young adults do. Mixing practice adds a second layer: in a JAES ducking study (ducking means lowering music or ambience while someone speaks), non-expert listeners preferred a loudness gap between commentary and background 4 LU larger, on average, than expert listeners did. LU, the loudness unit, moves step for step with decibels on a loudness meter. In short, the people who make the mix hear the dialogue more easily than the people who watch it.
| Metric | Value | Source |
|---|---|---|
| UK over-65s (offline sample) saying background music reduced their enjoyment of a TV programme | 62% | Hearing Health Matters 2011 |
| Programmes analysed in detail in the VLV, BBC and RNID audibility project | 21 | IPS, VLV Audibility of Speech on Television Project, 2011 |
| Recommended minimum loudness gap, commentary over music vs commentary over ambience (test with 22 normal-hearing listeners; non-experts preferred gaps 4 LU larger than experts) | at least 10 LU vs at least 15 LU | Torcoli et al., JAES 2019 |
| Difference in normalisation level for the same film when measured by overall loudness vs by speech | up to 14 dB | Lund and Skovenborg, SMPTE Annual Technical Conference 2014 |
| Speech reception threshold penalty, middle-aged (51 to 63) vs young (19 to 26) listeners, with music maskers | 0.4 to 1.6 dB | Baskent et al., JASA Express Letters 2014 |
| Same penalty with a competing talker vs with steady noise as the masker | 2.1 dB vs 0.8 dB | Baskent et al., JASA Express Letters 2014 |
| Dubbed movies (mainly Hollywood mainstream) analysed for the EBU loudness group’s cinematic-content work | 44 | AES Europe 2025 session, Camerer (ORF) |
The cinema factor has its own standards track. The EBU published R 128 s4, Loudness normalisation of Cinematic Content on November 8, 2023, introducing the loudness-to-dialogue ratio (LDR) because normalising action films to overall loudness can leave dialogue far below the -23 LUFS reference (AES Europe 2025 programme). The 14 dB spread above comes from the SMPTE paper, and the masking figures from Baskent et al.. Most recent available data for the masking and normalisation studies: 2014.
4. TV Speakers and the Playback Chain
A clean mix can still be ruined in the living room. Thin flat-panel cabinets force miniature drivers, and small, cheap loudspeakers in flat TVs can show level drops of about 30 dB per octave below 200 Hz (Astolfi et al., IEEE Transactions on Broadcasting 2021). The Politecnico di Torino and Rai team then showed that a filter applied by the broadcaster, with no new hardware at home, lifted perceived quality on three commercial sets.
The newest controlled study complicates the “buy more speakers” advice. Testing 60 adults, a 2026 JSLHR paper found that every group rated speech better at the more favourable dialogue-to-background ratio, but listeners with hearing loss rated five-channel surround among the poorest setups and plain stereo among the best (Ricketts et al., JSLHR 2026). Seeing faces matters as well, since audiovisual clips were part of the design; our lip-reading and visual speech statistics cover that cue. The AES responded at industry level with technical document TD1009, Improving Dialogue Intelligibility in Media, developed over more than two years and aimed partly at TV sets, set-top boxes and receivers (SVG, January 7, 2026).
| Metric | Value | Source |
|---|---|---|
| Low-frequency level drop of small flat-TV loudspeakers below 200 Hz | about 30 dB per octave | Astolfi et al., IEEE Transactions on Broadcasting 2021 (citing Behrends et al.) |
| Frequency band identified as most important for speech intelligibility | 0.5 kHz to 4 kHz | Astolfi et al., IEEE Transactions on Broadcasting 2021 |
| Volunteer listeners in the flat-TV test, across three commercial TV models | 72 | Astolfi et al., IEEE Transactions on Broadcasting 2021 |
| Average perceived quality gain from the broadcaster-side speech filter (standard deviation 14.6%) | 25.3% | Astolfi et al., IEEE Transactions on Broadcasting 2021 |
| Smallest gain, on the TV model with a sound pressure dip near 5 kHz | 14% | Astolfi et al., IEEE Transactions on Broadcasting 2021 |
| Participants in the speaker-configuration study (younger normal hearing, older normal hearing, older with hearing loss) | 3 groups of 20 | Ricketts et al., JSLHR 2026 |
| Dialogue-to-background ratios tested | about +7 dB and +12 dB | Ricketts et al., JSLHR 2026 |
| WDR participants with no clear preference who said the clear-speech and original mixes sounded too similar, by device: laptop speakers / TV speakers / headphones | 51% / 28% / 29% | Torcoli et al., IBC 2021 |
Statistical note: in the JSLHR study, audio-only speech understanding showed a group main effect, F(2, 50) = 63.50, p < .0001, and listeners with hearing loss rated surround lower than stereo (estimated marginal mean difference = 7.43, p = .01).
5. Dialogue Enhancement: What the Trials Measured
Object-based audio sends dialogue, music and effects as separate streams, so the TV or app can rebalance them for each viewer. The two largest public trials point the same way. In the BBC Casualty Accessible and Enhanced (A&E) Audio trial, 6,228 people tried a single slider between the broadcast mix and an accessible mix, and 83.6% of respondents said it made a difference (BBC R&D 2020). In Germany, the WDR and BR tests from September 2020 to April 2021 used deep-learning-based dialogue separation on existing mixes, which matters for programmes that were never produced as separate objects.
The preference data is more nuanced than the approval data. Only 46% of WDR participants preferred the clearer Dialog+ mix outright and 21% preferred the original, mainly because it sounded more natural (76% of that group). The appetite is concentrated where section 1 found the need: 62% of over-60s preferred Dialog+. A lab study with older listeners shows the scale of the gap, with chosen dialogue-to-background differences far above the default mixes, and a pupil-size study run in a VR headset found that dialogue enhancement, including the version built by source separation, reduced listening effort in 28 normal-hearing participants who heard 30 TV excerpts (Torcoli et al., QoMEX 2022, arXiv). Personalised mixes sit alongside other access services such as audio description on streaming and TV.
| Metric | Value | Source |
|---|---|---|
| People who tried the BBC Casualty accessible-audio slider vs respondents saying it made a difference | 6,228 vs 83.6% | BBC R&D, Creating Jade’s World, 2020 |
| Casualty respondents saying the slider made the content more enjoyable or easier to understand | 73.7% | BBC R&D, 2020 |
| WDR participants who liked having the option to switch to Dialog+: all / under 40 / over 60 | 83% / 76% / 88% | Torcoli et al., IBC 2021 |
| WDR participants preferring Dialog+ vs the original mix; Dialog+ share among over-60s | 46% vs 21%; 62% | Torcoli et al., IBC 2021 |
| Listeners aged 66 to 75 (n = 11): preferred speech-to-background loudness gap with original objects vs with dialogue separation vs default mix | 15.3 LU vs 13.5 LU vs 8.4 LU | Straninger, arXiv 2020 thesis study |
| Wimbledon NetMix experiment 2011: range by which listeners could raise or lower the commentary vs the standard mix | plus or minus 12 dB | EBU Technical Review, Fuchs, Tuff and Bustad, 2012 |
| NetMix survey respondents (98 respondents, about 1,200 player downloads) who agreed such technology would benefit TV vs radio | 84% vs 72% | EBU Technical Review, 2012 |
| Soundbar study participants who preferred speech enhancement switched off (vendor’s own study, run with RNID) | 15 of 215 | Sonos tech blog, June 10, 2025 (vendor claim) |
Context note: the Sonos figures are a manufacturer’s own study of its product and are shown for comparison only; the BBC and WDR trials are the independent benchmarks. The older-listener lab data is in the object-based audio study of people above 65, and the 2011 NetMix experiment in the EBU Technical Review (most recent available data for radio-to-TV comparisons: 2012).
6. Hearing Loss Prevalence as the Underlying Driver
Production habits explain why dialogue got harder; demographics explain why complaints keep rising. Over 25% of people older than 60 worldwide have disabling hearing loss (WHO fact sheet, March 3, 2026), and those are exactly the age groups that section 1 found struggling most. Hearing loss also changes how the set is used: in a Japanese study of older adults, preferred TV volume rose 3.4 dB for every 10 dB rise in average hearing level (Kurakata et al., Japanese Journal of Ergonomics 1999), so a shared living room rarely has one volume that suits everyone.
The treatment gap widens the problem. RNID estimates that over 10 million people in the UK could benefit from hearing aids but only 3 million use them (RNID prevalence data), and in the US fewer than 1 in 3 adults aged 70 and older who could benefit have ever used them (NIDCD Quick Statistics). An EBU review estimated that some 16% of Europeans have a hearing loss calling for treatment (EBU Technical Review 2012). For the progression curve by age, see our presbycusis progression statistics. This section is factual context only, not medical advice.
| Metric | Value | Source |
|---|---|---|
| People worldwide needing rehabilitation for disabling hearing loss | 430 million (over 5% of the world population) | WHO fact sheet, March 3, 2026 |
| People older than 60 affected by disabling hearing loss | over 25% | WHO fact sheet, March 3, 2026 |
| People projected to have some degree of hearing loss by 2050 | nearly 2.5 billion | WHO fact sheet, March 3, 2026 |
| People in the UK who are deaf, have hearing loss or tinnitus | over 18 million (one in three) | RNID, page updated October 21, 2026 |
| UK people with hearing loss: over 60 vs aged 70 or over | 65% vs almost 80% | RNID, 2026 |
| UK people who could benefit from hearing aids vs people who use them | over 10 million vs 3 million | RNID, 2026 |
| US adults with disabling hearing loss: aged 65 to 74 vs 75 and older | 22% vs 55% | NIDCD Quick Statistics, updated September 20, 2024 |
| Increase in preferred TV volume per 10 dB rise in average hearing level (older adults, Japan) | 3.4 dB | Kurakata et al., Japanese Journal of Ergonomics 1999 |
Outlier note: RNID’s UK estimates are modelled from the 1995 National Study of Hearing combined with census data and population projections, so they are projections rather than a fresh audiometric survey.
Summary: TV Dialogue Intelligibility Complaints by the Numbers
| Metric | Value | Source |
|---|---|---|
| WDR survey participants over 60 who often or very often struggle with TV speech | 90% | Torcoli et al., IBC 2021 |
| UK online panel adults vs offline over-65s reporting problems hearing TV speech | 70% vs 76% | Hearing Health Matters 2011 |
| UK adults who cannot always understand clearly what is said on TV | 71% | Hearing Health Matters 2011 |
| Viewers reporting difficulty with on-screen dialogue (vendor survey) | 84% | Xperi (DTS) own survey |
| Jamaica Inn complaints after episode one vs episode two | 546 vs 252 | Press Association, April 24, 2014 |
| SS-GB audience, episode one vs episode two | 6.1 million vs 3.9 million | Jewish News, February 28, 2017 |
| UK over-65s saying background music reduced their enjoyment | 62% | Hearing Health Matters 2011 |
| Speech reception threshold penalty for middle-aged vs young listeners, music maskers | 0.4 to 1.6 dB | Baskent et al., JASA Express Letters 2014 |
| Normalisation gap for the same film, loudness vs speech measure | up to 14 dB | Lund and Skovenborg, SMPTE 2014 |
| Low-frequency drop of small flat-TV loudspeakers below 200 Hz | about 30 dB per octave | Astolfi et al., IEEE Transactions on Broadcasting 2021 |
| Average perceived quality gain from a broadcaster-side speech filter | 25.3% | Astolfi et al., IEEE Transactions on Broadcasting 2021 |
| Dialogue-to-background ratios tested in the 2026 speaker study | about +7 dB and +12 dB | Ricketts et al., JSLHR 2026 |
| Casualty respondents saying the slider made content more enjoyable or easier to understand | 73.7% | BBC R&D 2020 |
| WDR participants who liked the option to switch to Dialog+ | 83% | Torcoli et al., IBC 2021 |
| People older than 60 with disabling hearing loss worldwide | over 25% | WHO fact sheet, March 3, 2026 |
| People needing rehabilitation for disabling hearing loss worldwide | 430 million | WHO fact sheet, March 3, 2026 |
| People in the UK who are deaf, have hearing loss or tinnitus | over 18 million | RNID 2026 |
| US adults 75 and older with disabling hearing loss | 55% | NIDCD 2024 |
Methodology and Sources
- Torcoli et al., Dialog+ in Broadcasting: First Field Tests Using Deep-Learning-Based Dialogue Enhancement, IBC 2021 (Fraunhofer IIS, WDR, BR), 2,126 survey participants - link
- Hearing Health Matters, What their telly tells them, most Brits can’t always tell, November 9, 2011 (reporting the BBC and Widex/Channel 4 audibility studies) - link
- Institute of Professional Sound, VLV’s Audibility of Speech on Television Project, March 2011 - link
- Xperi (DTS), Clear Audio: Understanding Audience Expectations for Dialogue Intelligibility (vendor-commissioned) - link
- YouGov, Most young people now watch TV and movies with the subtitles on, October 26, 2026 - link
- Press Association via WhatToWatch, Jamaica Inn draws more complaints despite BBC boosting sound levels, April 24, 2014 - link
- SVG (reporting PSNEurope), After Jamaica Inn and Mumblegate, IPS calls for better audibility in TV sound, September 25, 2014 - link
- The Telegraph via Yahoo News, SS-GB: BBC to re-examine sound after yet more mumbling complaints, February 20, 2017 - link
- Jewish News, BBC drama SS-GB loses a third of its viewers over sound complaints, February 28, 2017 - link
- BBC Academy, Why is loudness such an issue for TV viewers? - link
- Torcoli, Freke-Morin, Paulus, Simon and Shirley, Preferred Levels for Background Ducking to Produce Esthetically Pleasing Audio for TV with Clear Speech, JAES 67(12), 2019 - link
- Lund and Skovenborg, Loudness vs. Speech Normalization in Film and Drama for Broadcast, SMPTE 2014 Annual Technical Conference - link
- Baskent, van Engelshoven and Galvin, Susceptibility to interference by music and speech maskers in middle-aged adults, JASA 135(3), 2014 - link
- EBU, R 128 s4, Loudness normalisation of Cinematic Content, November 8, 2023 - link
- AES Europe 2025, Loudness of movies for broadcasting (Camerer, ORF), May 23, 2025 - link
- Astolfi et al., Speech Quality Improvement of Commercial Flat Screen TV-Sets, IEEE Transactions on Broadcasting 67(3), 2021 - link
- Ricketts et al., How Speaker Configuration of Recorded Media Affects Subjective Ratings for Listeners With and Without Hearing Loss, JSLHR 69(8), 2026 - link
- SVG, AES TD1009 Improving Dialogue Intelligibility in Media, January 7, 2026 - link
- BBC R&D, Creating Jade’s World (Casualty A&E Audio results), 2020 - link
- Straninger, Dialogue Enhancement in Object-based Audio: Evaluating the Benefit on People above 65, arXiv 2020 - link
- Torcoli, Robotham and Habets, Dialogue Enhancement and Listening Effort in Broadcast Audio: A Multimodal Evaluation, QoMEX 2022 - link
- Fuchs, Tuff and Bustad, Dialogue Enhancement: technology and experiments, EBU Technical Review 2012 - link
- Sonos tech blog, Arc Ultra speech enhancement study with RNID, June 10, 2025 (vendor’s own study) - link
- World Health Organization, Deafness and hearing loss fact sheet, March 3, 2026 - link
- RNID, Prevalence of deafness and hearing loss, updated October 21, 2026 - link
- NIDCD, Quick Statistics About Hearing, updated September 20, 2024 - link
- Kurakata et al., Relationship between hearing levels of the elderly and preferred volume levels of television, Japanese Journal of Ergonomics 35(3), 1999 - link
- Data watch: Jamaica Inn complaint totals conflict. Press Association counted 546 plus 252 after two episodes, a PSNEurope account based on BBC figures gave more than 1,000 over three nights, The Telegraph later wrote “more than 2,000”, and the widely repeated 2,200 figure could not be verified against a BBC primary page during this research, so we report only the per-episode counts as headline numbers. For SS-GB, The Telegraph described “dozens” of complaints, fewer than 100 by the Monday afternoon, and no final total was published in the sources we could read. The BBC Casualty trial counts 6,228 people who tried the slider, while the Dialog+ paper describes the same trial’s survey as having 299 participants, so the 83.6% and 73.7% shares refer to survey respondents, not everyone who tried it. The 2010 UK offline sample is given as 508 by Hearing Health Matters and as 506 effective diaries by the IPS project summary. The older-listener thesis prints 81.7% for 9 of 11 participants reporting TV speech problems (9 of 11 is 81.8%), so we do not use that share. Xperi and Sonos figures are vendor studies without published full methodology. Ofcom’s access-services research pages could not be retrieved during this update and are not cited. Several figures are most recent available data (BBC audibility study 2010, Jamaica Inn 2014, SMPTE and JASA 2014, EBU Technical Review 2012, Kurakata 1999); the AES TD1009 guidance and EBU R 128 s4 adoption are where newer measured data is most likely to appear.
Last updated: November 8, 2026. We update this roundup quarterly, and the next refresh is expected when the WHO next revises its deafness and hearing loss fact sheet or the AES publishes follow-up data on TD1009, Improving Dialogue Intelligibility in Media.