The frequency-following response, or FFR, is a valuable tool for examining the neural mechanisms involved in speech encoding. Due to its ability to closely mimic the temporal and spectral features of the eliciting auditory stimulus, the FFR represents a distinct subtype of auditory evoked potential (AEP). It reflects synchronized neural phase-locking to the spectro-temporal components of the acoustic signal (Krizman and Kraus, 2019; Coffey et al., 2019). While the FFR originates from multiple subcortical and cortical sources, it is believed to be primarily generated in the auditory midbrain, which is a hub of afferent and efferent activity (Coffey et al., 2019; Bidelman, 2015; Chandrasekaran and Kraus, 2010; Sohmer et al., 1977; White-Schwoch et al., 2017; White-Schwoch et al., 2019; Gorina-Careta et al., 2021). However, recent evidence also points to additional contributions from cortical areas, particularly for lower-frequency stimulus features (Coffey et al., 2016; Tichko and Skoe, 2017; Lerud et al., 2023; Bidelman, 2018). These attributes make the FFR particularly useful for studying central auditory processing of complex stimuli, such as speech, by providing insights into how neural networks encode acoustic features, thereby enhancing our understanding of speech perception (Kraus and White-Schwoch, 2015).

The FFR is shaped by various external factors, including listening context, challenging listening conditions (e.g., noise environments), age, and speech or language impairments (Gorina-Careta et al., 2022). Moreover, past auditory experiences significantly influence the encoding of a sound’s fundamental frequency (F0) in the FFR (Musacchia et al., 2007). For instance, language experience and bilingualism contribute to variations in the FFR (Krishnan et al., 2005; Anderson et al., 2013; Krizman et al., 2012). Moreover, several clinical conditions such as dyslexia (Chandrasekaran et al., 2009; Billiet and Bellis, 2011), mild cognitive impairment (Bidelman et al., 2017), and autism (Font-Alaminos et al., 2020; Otto-Meyer et al., 2018) have been shown to affect the FFR. Emerging evidence also suggests that the FFR could serve as a potential biomarker for auditory disorders like cochlear synaptopathy (Vasilkov et al., 2021; Verhulst et al., 2018) and auditory neuropathy (White-Schwoch et al., 2022). While earlier studies suggested that musical training enhances neural representations of speech and improves speech intelligibility (Weiss and Bidelman, 2015; Fleming et al., 2019), more recent large-scale investigations have challenged this view, finding no consistent evidence of enhanced FFRs in musically trained individuals when controlling for confounding variables (Whiteford et al., 2024; Riegel et al., 2024). These mixed findings indicate that the relationship between musical training and subcortical sound encoding may be more complex than previously assumed. Overall, these findings highlight the FFR as a dynamic measure, modulated by both external experiences and neurological conditions, providing valuable insights into auditory and cognitive processing.

The characteristics of the FFR in individuals with sensorineural hearing loss (SNHL) remain less understood. Our recent systematic review (Jacxsens et al., 2024) suggested a tendency towards prolongation of peak latencies in the time domain with increasing hearing loss, potentially indicating delays in neural processing of auditory stimuli. Additionally, several studies reported a reduction in the F0 response amplitude in individuals with SNHL compared to normal-hearing controls, suggesting deficits in encoding pitch information. However, results regarding encoding of the temporal fine structure and FFR peak amplitudes were inconclusive. We also highlighted a large variability in participant characteristics, stimulus parameters, and FFR outcome parameters across studies. Many studies included older SNHL participants than their normal-hearing counterparts, and stimulus durations ranged 40 to 543 ms. Since longer stimuli allow for better phase locking, a longer duration of the sustained segment of the stimulus elicits sustained subcortical responses reflecting synchronous neural phase locking, which are reflected in the FFR (Skoe and Kraus, 2010; Kraus et al., 2017). Consequently, stimuli need to have a sufficient length to allow this phase locking to occur. Therefore, we recommended stimuli of at least 170 ms. Another notable limitation in many studies was the lack of control for audibility, as most presented stimuli at equal intensity (dB SPL) rather than accounting for sensation level (dB SL), making it difficult to disentangle effects of SNHL from those of reduced stimulus audibility. Given these findings, further research is needed to clarify these factors, particularly under experimental conditions which are controlled as much as possible.

The /da/ syllable is the most commonly used FFR stimulus. In our systematic review, 10 of the 16 included studies used /da/ (Jacxsens et al., 2024), similar to Lemos et al. (Lemos et al., 2021), where 8 of 15 studies on newborns and infants used it. Its popularity stems from its universality in European languages (Maddieson, 1984) and its combination of transient and sustained periodic components, with the /d/ onset resembling the click ABR and the /a/ providing a sustained periodic portion (Skoe and Kraus, 2010). This stimulus elicits clear, robust responses and is relevant for studying populations with hearing or learning impairments, as stop consonants are perceptually challenging, especially in noise (Kraus et al., 1996; Turner et al., 1992). However, this stimulus is less effective for evaluating the temporal fine structure due to its short transition between /d/ and /a/ and because the higher-frequency formants (>500 Hz) it contains are beyond the optimal range for neural phase-locking (Gorina-Careta et al., 2022; Arenillas-Alcón et al., 2021).

The diphthong /oa/ was recently introduced as an alternative FFR stimulus, particularly for infants (Arenillas-Alcón et al., 2021; Arenillas-Alcón et al., 2023; Gorina-Careta et al., 2024). This stimulus features two vowel sections with lower-frequency harmonics, allowing better assessment of F0 and temporal fine structure (Gorina-Careta et al., 2022; Arenillas-Alcón et al., 2021). However, this stimulus also has its limitations. Arenillas-Alcón et al. (Arenillas-Alcón et al., 2023) found that newborns showed a smaller F0 response amplitude to /a/ when preceded by /o/ than by /d/ in /da/.

Building on this, we conducted a multi-site stimulus optimization study involving 34 normal-hearing adults (18–40 years) at Antwerp University Hospital and the University of Barcelona (Jacxsens et al., 2025). Four different stimuli (/dao/, /doa/, /ao/, and /oa/) designed and equated for loudness were presented in three blocks of 1000 trials in randomized block order. The /ao/ stimulus elicited the largest F0 response amplitude, while the /oa/ stimulus elicited the smallest F0 response amplitude. The F1 response amplitude displayed a different pattern, with the /oa/ stimulus resulting in the largest amplitudes for both the /o/ and /a/ sections of the stimulus. Our findings suggest that the F0 and F1 response amplitudes are influenced by the frequency structure of the eliciting vowels and by the preceding vowels and consonants. The choice of stimulus should align with the study’s goals. The /ao/ stimulus is ideal for studies focusing on the frequency-following response to the envelope (FFRENV), the /oa/ stimulus for studies focusing on the frequency-following response to the temporal fine structure (FFRTFS). We recommended using the /dao/ stimulus for studies in which both the FFRENV and FFRTFS are examined, since this stimulus offers strong responses for both the FFRENV and the FFRTFS and allows comparison with previous studies using the /da/ stimulus.

In this study, we further investigate the impact of the severity of SNHL on the FFR, addressing previous limitations with a larger patient cohort and using the optimized /dao/ stimulus.