Revisiting expectations for average and soft speech levels

More than 40 years later, is 65 dBA still an accurate estimate for average speech levels? What level is soft speech?

When prescribing hearing aids, our first goal is typically restoration of speech audibility while accommodating a patient’s dynamic range. In this process, many assumptions are based on the expectation that the level of average conversational speech varies around 65 dBA.

One of the most frequently cited resources on average speech levels was published by Pearsons, Bennett, and Fidell (1977), who collected these data as part of a report to the U.S. Environmental Protection Agency. In this report, the authors summarized the results of acoustic measurements in a wide variety of environments that included homes, schools, and workplaces.

Scope and findings on speech levels

Looking back, the scope of this work is impressive. The authors collected information in more than 20 classrooms, 25 homes, 4 hospitals, 7 stores, and 4 commercial aircraft with passengers onboard. The distance for recording speech levels varied across most locations, but most measurements were collected at or normalized to a 1-meter distance from the talker. To this day, we see that many research designs continue to use 1-meter as a standard talker-to-listener distance.

The data showed that average conversational levels outdoors were 65 dBA while speech levels inside a typical home were 55 dBA. In comparison, average speech levels in stores were 61 dBA and 77 dBA in commercial aircraft. Based on our convention of 65 dBA, it may seem surprising that conversational levels in quiet (home) environments were recorded at 10 dB lower more than 40 years ago.

Impact of background noise

It’s important to consider that real-life environments contain background noise that directly affects speech production. As this was known to the authors, they also recorded speaking levels from 100 different talkers in a quiet anechoic chamber. The levels recorded in the anechoic environment ranged from 50 dBA for casual or relaxed speaking to 89 dBA for shouted speaking; normal speaking effort was close to 58 dBA, which is 6-7 dB below the 65 dBA reference. However, the addition of background noise at ecologically valid levels elevated those averages toward the expected range, a finding that was confirmed by Sato and colleagues in 2011.

Challenges in measuring soft speech

The fact that average speech levels vary significantly and have a dependency on background noise levels prompts one to wonder about the levels of quiet or soft speech. Interestingly, measurement of soft speech is challenging, as many acoustic cues are obscured by background noise or the noise floor of the recording equipment.

Šrámková and colleagues (2015) addressed this challenge by measuring the lowest level components of speech with modern recording equipment and sound processing techniques that ensured sound level estimates were only taken from speech components and not the silence or noise between speech utterances.

Through this systematic process, they determined that the softest human vocal sounds fall between 42 and 47 dBA. While these levels are quite low, they were recorded in quiet and at a distance of 30 centimeters (1 foot) from the talker’s mouth. If we translate these findings to real-world scenarios, the effective levels of soft speech would be even lower with the effects of background noise and increased distances between the talker and listener. This teaches us that quiet speech levels are very low, and restoring audibility of these cues may be challenging.

Implications for hearing aid prescription

When we consider this information in the prescription of hearing aids, there is good rationale for the use of 65 dBA as an average level for conversational speech in some background noise. Soft speech sounds are, however, extremely quiet which makes them susceptible to masking by even low levels of background noise.

Verification of gains for low-level sounds is one step toward ensuring audibility of speech for our patients, a second step is considering adaptive technologies designed to selectively improve audibility of low-level speech. As an example, Phonak’s Speech Enhancer was developed to identify and increase audibility of soft-speech cues resulting in significant clinical benefits (Latzel et al., 2023; Manning et al., 2023).

While many clinical assumptions stand the test of time, there are aspects that change with the steady evolution of technology. In this example, we see that estimates of real-world conversations may indeed fall close to 65 dBA. Yet the softest components of speech are even lower in level than we routinely consider. This helps to reinforce the importance of verifying low-level speech audibility and familiarizing ourselves with the benefits of hearing aid features designed to improve access to these delicate conversational cues.

To gain deeper insights into how Phonak’s Speech Enhancer can improve hearing aid performance, check out our previous blog article: Speech Enhancer: 3 Real-World Problems It Solves.


  1. Pearsons, K. S., Bennett, R. L., & Fidell, S. A. (1977). Speech levels in various noise environments. Office of Health and Ecological Effects, Office of Research and Development, US EPA.
  2. Sato, H., Morimoto, M., & Ota, R. (2011). Acceptable range of speech level in noisy sound fields for young adults and elderly persons. The Journal of the Acoustical Society of America130(3), 1411-1419.
  3. Šrámková, H., Granqvist, S., Herbst, C. T., & Švec, J. G. (2015). The softest sound levels of the human voice in normal subjects. The Journal of the Acoustical Society of America137(1), 407-418.
  4. Latzel, M., Lesimple, C., & Woodward, J. (2023). Speech Enhancer significantly reduces listening effort and increases intelligibility for speech from a distance. Phonak Field Study News.  Retrieved from
  5. Manning, J., Wolfe, J., Neumann, S., Nelson, J., & Dunn, A. (2023, May 11-13). The use of Speech Enhancer with pediatric hearing aid users [Conference poster]. 7th Phonak European Conference, Berlin, Germany. Retrieved from