pith. sign in

arxiv: 2606.17263 · v1 · pith:O6UFOZN5new · submitted 2026-06-15 · 📡 eess.AS

Direction of arrival estimation from distant microphone data using single frequency filtering

Sushmita Thakallapalli , Sudarsana Reddy Kadiri , Nilesh Madhu , Suryakanth V Gangashetty This is my paper

Pith reviewed 2026-06-27 02:12 UTC · model grok-4.3

classification 📡 eess.AS
keywords direction of arrival estimationsingle frequency filteringnarrowband methodsspatial aliasingdistant microphonesmultiple speakersspeech processingcross-correlation
0
0 comments X

The pith

Single frequency filtering makes narrowband DoA estimation robust to spatial aliasing while preserving multi-speaker resolution.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper develops a narrowband direction-of-arrival estimator that applies single frequency filtering to microphone signals and then performs cross-correlation only on the resulting speech-present time-frequency regions. This targets the problem that standard narrowband methods suffer from spatial aliasing in distant-microphone recordings, while broadband methods lose the ability to resolve multiple speakers within one time frame. The authors argue that SFF produces high-SNR regions where speech versus non-speech decisions stay reliable even under reverberation and noise, allowing the narrowband estimator to gain robustness without sacrificing its sparsity advantage. Results on simulated and real data show the new narrowband method exceeds the prior narrowband baseline and surpasses some broadband estimators across all tested conditions.

Core claim

The proposed SFF-based narrowband estimator, which computes DoAs from cross-correlations of speech-present time-frequency regions identified in the single frequency filtering domain, outperforms the state-of-the-art narrowband estimator and performs better than some broadband estimators in detection and accuracy metrics on both simulated and real-world data under varying reverberation and noise.

What carries the argument

Single frequency filtering spectrum whose high-SNR speech-present time-frequency regions are isolated for cross-correlation-based narrowband DoA estimation.

If this is right

  • The SFF-based narrowband estimator outperforms the state-of-the-art narrowband approach on detection and accuracy metrics in every tested environment.
  • Its accuracy exceeds that of some broadband estimators while still resolving multiple speakers inside a single time frame.
  • The method applies equally to simulated and real distant-microphone recordings across different reverberation and noise levels.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same high-SNR regions identified by SFF could be reused for joint DoA and source separation tasks.
  • Parameter tuning of the SFF pole radius might further improve aliasing resistance in specific room geometries.
  • The approach suggests a route to hybrid estimators that switch between narrowband and broadband modes based on detected source count.

Load-bearing premise

The SFF domain continues to provide regions of high signal-to-noise ratio and reliable speech versus non-speech discrimination even when the microphone signals are degraded by distance, reverberation, and noise.

What would settle it

A direct comparison on the same distant-microphone recordings in which the SFF-based narrowband method records lower detection rates or larger angular errors than the state-of-the-art narrowband baseline would falsify the performance claim.

Figures

Figures reproduced from arXiv: 2606.17263 by Nilesh Madhu, Sudarsana Reddy Kadiri, Suryakanth V Gangashetty, Sushmita Thakallapalli.

Figure 1
Figure 1. Figure 1: Block diagram of the computation of the spectral envelope [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: TF representations for a synthetic signal consisting of impulses and narrowband com [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: An example of computing spectral flatness for a sentence. Panel (a) shows the speech [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
read the original abstract

In distant microphones, broadband (BB) methods for direction-of-arrival (DoA) estimation are more suitable than narrowband (NB) methods. Due to the aggregation of their optimization function across all frequency bands, BB estimators are robust to spatial aliasing, a known problem in processing distant microphone data. In NB methods, DoA estimation is performed by utilizing \textit{local} information in each frequency band and hence the estimation is affected by spatial aliasing. However, unlike BB methods, NB methods exploit frequency sparsity to estimate the DoAs of \textit{multiple speakers} in a \textit{single time frame}. In this article, a method to improve the robustness of a NB DoA estimator to spatial aliasing is developed. The proposed method is based on cross-correlation of speech-present time-frequency regions obtained by single frequency filtering (SFF) of the microphone signals. The SFF spectrum is chosen because SFF components have regions of high signal-to-noise ratio both in time and frequency and because speech and non-speech discrimination is robust to degradations in the SFF domain. The proposed NB estimator is compared to four state-of-the-art estimators (one NB and three BB) using detection and accuracy metrics on simulated and real-world data in different reverberation and noise conditions. The results show that in all the environments, the SFF-based NB approach outperforms the state-of-the-art NB approach. Furthermore, the performance of the SFF-based approach is better than some of the BB estimators.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The manuscript proposes a single-frequency-filtering (SFF) based narrowband (NB) direction-of-arrival (DoA) estimator that identifies high-SNR speech-present time-frequency regions for cross-correlation. It claims this approach outperforms a state-of-the-art NB estimator and some broadband (BB) estimators on both simulated and real data under various reverberation and noise conditions, using detection and accuracy metrics.

Significance. The work addresses a practical limitation of NB DoA methods in distant microphone setups by improving robustness to spatial aliasing while preserving the ability to handle multiple speakers per frame. If the empirical gains hold under rigorous controls, it offers a useful hybrid approach between NB and BB methods.

major comments (2)
  1. [Abstract and §4] Abstract and §4 (results): the central claim of outperformance on detection and accuracy metrics across all simulated and real environments lacks reported numerical values, trial counts, statistical significance tests, or controls for array geometry and speaker count, making the empirical result unverifiable from the given description.
  2. [§3] §3 (method): the key assumption that SFF components provide robust high-SNR regions and speech/non-speech discrimination is stated without quantitative validation or ablation against the specific reverberation and noise degradations used in the experiments.
minor comments (1)
  1. [§3] Equations for SFF spectrum computation and the cross-correlation step should be numbered and referenced explicitly in the method description.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the detailed and constructive review. We address each major comment below and will revise the manuscript to improve verifiability and strengthen the methodological justification.

read point-by-point responses

  1. Referee: [Abstract and §4] Abstract and §4 (results): the central claim of outperformance on detection and accuracy metrics across all simulated and real environments lacks reported numerical values, trial counts, statistical significance tests, or controls for array geometry and speaker count, making the empirical result unverifiable from the given description.

    Authors: We agree that the abstract and §4 would benefit from explicit numerical values, trial counts, and clearer statements on controls. The experiments in §4 do control for array geometry (fixed 4-microphone linear array) and speaker count (up to 3 concurrent speakers per frame), with results aggregated over multiple simulated and real recordings; however, these details are only summarized rather than quantified in the text. In the revision we will insert representative numerical results (e.g., detection rates and angular errors) from the tables/figures, state the number of trials, and explicitly note the controls for geometry and speaker count. We did not perform formal statistical significance tests; if the referee considers them essential we can add them or note their absence as a limitation. revision: yes

  2. Referee: [§3] §3 (method): the key assumption that SFF components provide robust high-SNR regions and speech/non-speech discrimination is stated without quantitative validation or ablation against the specific reverberation and noise degradations used in the experiments.

    Authors: The claim draws on established properties of SFF reported in prior speech-processing literature, but we acknowledge that direct quantitative validation or ablation under the exact reverberation times and SNR conditions of our experiments is missing from §3. We will add a short subsection or supplementary figure that reports SNR histograms and speech/non-speech classification accuracy of SFF components on the same simulated and real data used in §4, thereby providing the requested validation. revision: yes

Circularity Check

0 steps flagged

No circularity: empirical method with external validation

full rationale

The paper describes an algorithmic NB DoA estimator that applies cross-correlation to SFF-derived speech-present TF regions, motivated by SNR and discrimination properties of SFF. Central claims are performance gains versus four comparator estimators (one NB, three BB) measured by detection/accuracy metrics on simulated and real data under varied reverberation/noise. No mathematical derivation chain exists that reduces to self-definition, fitted inputs renamed as predictions, or load-bearing self-citations. The method is presented as a practical improvement and is tested against independent baselines; results are falsifiable via replication on the same data conditions. This matches the default expectation of a non-circular empirical contribution.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only review provides no explicit free parameters, axioms, or invented entities; full paper would be needed to audit these.

pith-pipeline@v0.9.1-grok · 5824 in / 1087 out tokens · 45480 ms · 2026-06-27T02:12:31.084012+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

35 extracted references · 2 canonical work pages

  1. [1]

    L. Wang, T. Hon, J. D. Reiss, A. Cavallaro, Self-localization of ad-hoc arrays using time difference of arrivals, IEEE Trans. Signal Process. 64 (4) (2016) 1018–1033

    2016

  2. [2]

    L. Wang, T. Hon, J. D. Reiss, A. Cavallaro, An iterative approach to source counting and localization using two distant microphones, IEEE Trans. Au- dio, Speech and Lang. Process. 24 (6) (2016) 1079–1093

    2016

  3. [3]

    Knapp, G

    C. Knapp, G. Carter, The generalized correlation method for estimation of time delay, IEEE Trans. Acoustics, Speech and Signal Process. 24 (4) (1976) 320–327

    1976

  4. [4]

    Robust localization in re- verberant rooms

    L. H. DiBiase, H. F. Silverman, M. S. Brandstein, "Robust localization in re- verberant rooms" in Springer: Microphone Arrays - Signal Processing Tech- niques and Applications, 2001, pp. 157–180

    2001

  5. [5]

    Liu, B.C

    C. Liu, B.C. Wheeler, W.D. O’Brien, D.William, R.C. Bilger, C.R. Lans- ing and A.S. Feng, Localization of multiple sound sources with two micro- phones, J.Acoust. Soc. Am. 108 (4) (2000) 1888–1905

    2000

  6. [6]

    Christof, M

    F. Christof, M. Juha, Source localization in complex listening situations: Selection of binaural cues based on interaural coherence, J. Acoust. Soc. Am. 116 (5) (2004) 3075–3089

    2004

  7. [7]

    Mouba, H

    J. Mouba, H. Marchand, Source localization/separation/respatialization sys- tem based on unsupervised classification of interaural cues, in: Proc. Conf. Digital Audio Effects, 2006, pp. 233–238

    2006

  8. [8]

    Viste, G

    H. Viste, G. Evangelista, On the use of spatial cues to improve binaural source separation, in: Proc. Conf. Digital Audio Effects, 2003, pp. 209–213

    2003

  9. [9]

    Loesch, B

    B. Loesch, B. Yang, Source number estimation and clustering for under- determined blind source separation, in: Proc. IEEE Int. Workshop on Acous- tic Signal Enhancement (IW AENC), 2008

    2008

  10. [10]

    Madhu, R

    N. Madhu, R. Martin, A scalable framework for multiple speaker localiza- tion and tracking, in: Proc. IEEE International Workshop on Acoustic Signal Enhancement (IW AENC), 2008, pp. 1–4. 16

    2008

  11. [11]

    Thakallapalli, S

    S. Thakallapalli, S. V . Gangashetty, N. Madhu, NMF-weighted SRP for multi-speaker direction of arrival estimation: Robustness to spatial aliasing while exploiting sparsity in the atom-time domain, EURASIP J. Audio Speech Music. Process. (1) (2021) 13.doi:10.1186/ s13636-021-00201-y. URLhttps://doi.org/10.1186/s13636-021-00201-y

    work page doi:10.1186/s13636-021-00201-y 2021

  12. [12]

    Aneeja and B

    G. Aneeja and B. Yegnanarayana, Single frequency filtering approach for discriminating speech and nonspeech, IEEE Trans. Audio, Speech and Lan- guage Process. 23 (4) (2015) 705–717

    2015

  13. [13]

    Renevey, A

    P. Renevey, A. Drygajlo, Entropy Based V oiced Activity Detection in Very Noisy Conditions, in: Proc. Int. Conf. on Speech Communication and Tech- nology (INTERSPEECH), 2001

    2001

  14. [14]

    Madhu, Note on measures for spectral flatness, Electronics Letters 45 (23) (2009) 1195–1196

    N. Madhu, Note on measures for spectral flatness, Electronics Letters 45 (23) (2009) 1195–1196

    2009

  15. [15]

    V . C. Raykar, B. Yegnanarayana, S. R. M. Prasanna and R. Duraiswami, Speaker localization using excitation source information in speech, IEEE Trans. Speech and Audio Process. 13 (5) (2005) 751–761

    2005

  16. [16]

    S. R. Kadiri, B. Yegnanarayana, Epoch extraction from emotional speech using single frequency filtering approach, Speech Comm. 86 (2017) 52 – 63

    2017

  17. [17]

    S. R. Kadiri, B. Yegnanarayana, Determination of glottal closure instants from clean and telephone quality speech signals using single frequency fil- tering, Comput. Speech Lang. (2020) 101097

    2020

  18. [18]

    Aneeja, S

    G. Aneeja, S. R. Kadiri, B. Yegnanarayana, Detection of glottal closure in- stants in degraded speech using single frequency filtering analysis, in: Proc. Interspeech, 2018, pp. 2300–2304

    2018

  19. [19]

    Gurugubelli and A

    K. Gurugubelli and A. K. Vuppala, Perceptually enhanced single frequency filtering for dysarthric speech detection and intelligibility assessment, in: Proc. Int. Conf. Acoustics Speech and Signal Processing (ICASSP), 2019, pp. 6410–6414

    2019

  20. [20]

    Murthy, B

    N. Murthy, B. Yegnanarayana, and S. R. Kadiri, Time delay estimation from mixed multispeaker speech signals using single frequency filtering, Circuits, Systems, and Signal Processing. 17

  21. [21]

    Z. Wang, X. Zhang, D. Wang, Robust speaker localization guided by deep learning-based time-frequency masking, IEEE Trans. Audio, Speech and Language Process 27 (1) (2019) 178–188

    2019

  22. [22]

    H. Kang, M. Graczyk, J. Skoglund, On pre-filtering strategies for the GCC- PHAT algorithm, in: Proc. IEEE International Workshop on Acoustic Signal Enhancement (IW AENC), 2016, pp. 1–5

    2016

  23. [23]

    IEEE Workshop Appl

    J.Traa, P.Smaragdis, N.D.Stein, D.Wingate, Directional NMF for Joint Source Localization and Separation, in: Proc. IEEE Workshop Appl. Sig- nal Process. Audio Acoust., 2015, pp. 1–5

    2015

  24. [24]

    Thakallapalli, S

    S. Thakallapalli, S. R. Kadiri, S. V . Gangashetty, Spectral features derived from single frequency filter for multispeaker localization, in: 2020 National Conference on Communications (NCC), 2020, pp. 1–6

    2020

  25. [25]

    Chennupati, S

    N. Chennupati, S. R. Kadiri, B. Yegnanarayana, Significance of phase in single frequency filtering outputs of speech signals, Speech Communication 97 (2018) 66–72

    2018

  26. [26]

    E. A. P. Habets and S. Gannot, Generating sensor signals in isotropic noise fields, J. Acoust. Soc. Am. 122 (6) (2007) 3464–3470

    2007

  27. [27]

    Sivasankaran, Room simulator in python,https://github.com/ sunits/rir_simulator_python, [Last accessed: ]

    S. Sivasankaran, Room simulator in python,https://github.com/ sunits/rir_simulator_python, [Last accessed: ]

  28. [28]

    Liutkus, F.R

    A. Liutkus, F.R. Stöter, Z.Rafii, D.Kitamura, B.Rivet, N.Ito, N.Ono, and J.Fontecave, The 2016 signal separation evaluation campaign, in: Latent Variable Analysis and Signal Separation, Springer International Publishing, Cham, 2017, pp. 323–332

    2016

  29. [29]

    Varga, H

    A. Varga, H. J. M. Steeneken, Assessment for automatic speech recognition: II. NOISEX-92: A database and an experiment to study the effect of additive noise on speech recognition systems, Speech Comm. 12 (3) (1993) 247–251. URLhttp://www.speech.cs.cmu.edu/comp.speech/Section1/ Data/noisex.html

    1993

  30. [30]

    N. Ono, Z. Koldovský, S. Miyabe, N. Ito, The 2013 signal separation evalua- tion campaign, in: Proc. IEEE International Workshop on Machine Learning for Signal Processing (MLSP), 2013, pp. 1–6. 18

    2013

  31. [31]

    J. H. DiBiase, A high-accuracy, low-latency technique for talker localization in reverberant environments using microphone arrays, PhD thesis, Brown University, 2000

    2000

  32. [32]

    Makhoul, Linear prediction: A Tutorial Review, Proceedings of the IEEE 63 (4) (1975) 561–580.doi:10.1109/PROC.1975.9792

    J. Makhoul, Linear prediction: A tutorial review, Proc. IEEE 63 (4) (1975) 561–580.doi:10.1109/PROC.1975.9792

    work page doi:10.1109/proc.1975.9792 1975

  33. [33]

    Perez-Lorenzo, R

    J. Perez-Lorenzo, R. Viciana-Abad, P. Reche-Lopez, F. Rivas, J. Escolano, Evaluation of generalized cross-correlation methods for direction of arrival estimation using two microphones in real environments, Applied Acoustics 73 (8) (2012) 698 – 712

    2012

  34. [34]

    Zhang, D

    C. Zhang, D. Florencio, Z. Zhang, Why does PHAT work well in lownoise, reverberative environments?, in: Proc. Int. Conf. Acoustics Speech and Sig- nal Processing (ICASSP), 2008, pp. 2565–2568

    2008

  35. [35]

    Yegnanarayana, C

    B. Yegnanarayana, C. Avendano, H. Hermansky, P. S. Murthy, Speech en- hancement using linear prediction residual, Speech Comm. 28 (1) (1999) 25 – 42. 19

    1999