Next Article in Journal
Global Mobility Networks of Smart City Researchers: Spatiotemporal and Multi-Scale Perspectives, 2000–2020
Previous Article in Journal
Economic Assessment of Building Adaptation to Climate Change: A Systematic Review of Cost Evaluation Methods
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Smart Cities
Volume 8
Issue 5
10.3390/smartcities8050157
smartcities-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Automatic Speech Recognition of Public Safety Radio Communications for Interstate Incident Detection and Notification

by
Christopher M. Gartner
1,*,
Vihaan Vajpayee
2,
Jairaj Desai
1 and
Darcy M. Bullock
1
1
Lyles School of Civil and Construction Engineering, Purdue University, West Lafayette, IN 47907, USA
2
Department of Computer Science, Purdue University, West Lafayette, IN 47907, USA
*
Author to whom correspondence should be addressed.
Smart Cities 2025, 8(5), 157; https://doi.org/10.3390/smartcities8050157
Submission received: 6 August 2025 / Revised: 18 September 2025 / Accepted: 19 September 2025 / Published: 24 September 2025
Browse Figures
Versions Notes

Abstract

Highlights

What are the main findings?
  • Modern Automatic Speech Recognition models are capable of transcribing 9-1-1 dispatch communications with usable accuracy.
  • Commercial products for recording, transcribing, and disseminating information from 9-1-1 dispatch communications promptly for Traffic Incident Management.
What is the implication of the main finding?
  • Using existing products and software, Traffic Management Centers can expand their effective coverage to more rural segments by automatically relaying regional 9-1-1 dispatch broadcasts to a central source.

Abstract

Most urban areas have Traffic Management Centers that rely partially on communication with 9-1-1 centers for incident detection. This level of awareness is often lacking for rural interstates spanning several 9-1-1 centers. This paper presents a novel approach to extending TMC visibility by automatically monitoring regional 9-1-1 dispatch channels using off-the-shelf hardware and open-source speech-to-text libraries. Our study presents a proof-of-concept study servicing 71 miles of rural I-65 in Indiana, successfully monitoring four county dispatch centers from a single location, and efficiently transcribing live audio within 60 s of broadcast. This work’s primary contribution is demonstrating the feasibility and practical value of automated incident detection systems for rural interstates. This technology is implementation-ready for extending the visibility of Traffic Management Centers in rural interstate segments. Further work is underway for developing scalable procedures for integrating multiple remote sites, extracting more diverse keyword sets, investigating optimal speech-to-text models, and assessing the technical aspects of the experimental procedures of this manuscript.

1. Introduction

Traffic Management Centers (TMCs) and their associated Motorist Information Systems (MISs) rely on timely data to manage incidents. However, a significant challenge exists for rural areas, where the flow of near-real-time incident information for public safety dispatch centers to the central TMC is often delayed or nonexistent. While 9-1-1 centers have the most accurate and immediate information, their primary focus on emergency response often means that TMCs (and thus motorists) are notified only after the incident has caused traffic disruptions [1,2,3].
This manuscript presents a novel approach to bridge this information gap: relaying information from rural 9-1-1 dispatch centers to a central TMC (and thus the motorists) promptly. By integrating off-the-shelf equipment and open-source Automatic Speech recognition (ASR) libraries, we show the feasibility of transcribing dispatch audio to automatically identify and relay critical incident details to a central TMC. Similar applications of ASR have been explored in industries such as Air Traffic Control (ATC) and the medical field for dictation of patient notes; however, no application to Traffic Incident Management (TIM) has been made. This novel approach expands the geographical awareness of TMC operators by efficiently providing local first responder dispatch and response communication.
This study details an experimental setup used to evaluate the feasibility of monitoring several regional 9-1-1 dispatch centers, transcribing the spoken language using ASR, and using keywords to determine relevance. The successful outcomes of this feasibility study serve as a step toward strengthening Information and Communications Technology (ICT) within traffic management. The objective of this approach is to provide faster incident detection, reduce agency time to provide timely communication to motorists via highway message boards, motorist information systems, and, in some cases, press detours and press releases.

1.1. Problem Statement

Although there has been extensive work in the area of incident detection, this paper explores the feasibility of using Automatic Speech Recognition of public safety emergency dispatch channels to identify keywords such as “interstate,” “mile marker,” “fire,” “crash,” and several other related words to “detect” potential interstate incidents and their approximate location.

1.2. Document Overview

This paper is organized as follows. Section 2 reviews the relevant literature, providing the foundational knowledge for this study. Section 2.6 introduces the study’s location and presents a case study that underscores the critical nature of this research. Section 4 details the core methodologies developed and implemented, including the processes for audio capture, transcription, keyword extraction, and automated notification. Section 5 presents the results of our study, followed by a detailed analysis of the case study in Section 6. Finally, the maunscript discusses the proposed implementation needs, outlines the study’s limitations and future scope, and Section 7 offers concluding remarks.

2. Background

2.1. Incident Detection Literature

First implementations of incident detection utilized a combination of patrolling motorists (e.g., police, mechanics, etc.), citizen’s band (CB) radio in private and transit vehicles [4], closed-circuit television, aerial surveillance, emergency call boxes, and cooperative motorist aid systems [5,6].
Early systematic adopters favored inductive loop detectors (ILDs) for their “unmanned” nature [7]. Advancements in analyzing traffic patterns for abnormalities between stations became the standard method with advancements in algorithms, internet connection, and robust units [4,8,9,10,11,12,13,14]. In the 2000s, with advancements in computing power, machine-based detection methodologies like computer vision, radar, and LiDAR detection, environmental audio cues, and smartphone-based methods emerged as alternatives to ILDs [12,15,16,17,18,19,20].
One of the most recent additions to this list of methodologies is decentralized incident detection and notification [17], which leverages vehicle-to-everything communications to interface with other vehicles, infrastructure, and other sources of information. Crowdsourced options have also grown, relaying information between drivers using navigation applications [21].
Abroad, “Automatic Incident Detection” (AID) is widely researched. In recent years, Machine Learning (ML) based AI has become increasingly popular. ML approaches include video-based systems using CCTV networks, comparative analysis between road segment speeds, and further work on inductive loop systems [22,23,24].
Despite the multitude of methodologies, the communication and inference required to effectively use any one of them still leaves the requirement of communicating with the detection personnel and operators. The following sections will discuss the methods used to capture and relay information to TMC operators, and the interdisciplinary technology that has the potential to fill this void.

2.2. Public Safety Radio Communications

Historically, public safety dispatch has been carried out over a shared frequency. As the radio spectrum has become increasingly congested, public safety agencies are increasingly incorporating trunked radio communications in the 800 MHz band that can be dynamically shared [25]. In this study, both traditional trunked frequencies maintained by the state of Indiana (SAFE-T) and Tippecanoe County were monitored [26,27,28,29,30]. The voice quality of modern digital radio systems is comparable to that of modern cellular communications, whereas traditional systems exhibit more variations in quality. Sites such as RadioReference.com serve as a crucial source of information, acting as a crowd-sourcing agency for frequency assignments and digital identifiers necessary to monitor both types of systems [31,32].
The methods used in this study are primarily focused on standards in the United States, primarily around Project 25 or P25. For relevance abroad, this is comparable to the European “Terrestrial Trunked Radio” standards or TETRA [33].

2.3. Automatic Speech Recognition

Automatic Speech Recognition (ASR) is not a new field, but a still-developing field in computer science that aims to transform spoken language into text using computers. Early systems were developed in the 1980s with limited results, but began to advance rapidly in the mid-1990s [34,35]. By the turn of the century, major brands were implementing ASR engines within several of their projects, and by the 2010s, it had become a staple in many industries (e.g., health, telecommunications, education, and home automation) [36].
The combination of high-quality, digital trunked systems and ASR engines is now enabling real-time high-quality speech-to-text (STT) transcription. The quality of STT transcription is quantified using a metric called “Word Error Rate” (WER), which gives the accuracy of a model’s output when compared to a ground truth [37,38].

2.4. Traffic Management Centers and Traffic Incident Management

TMCs have existed since the early 1970s [39]. In many ways, they have not changed their primary objective and toolkit: monitor traffic flows, coordinate traffic incident management, assist first responders, and provide timely information to motorists [39,40]. TMCs utilize a combination of 9-1-1 calls, first responder calls, motorist calls, traffic speed monitoring equipment, and vast Intelligent Transportation System (ITS) camera networks to perform their tasks.

2.5. Related Work: Integrating Public Safety Radio with Automatic Speech Recognition

In the published literature, a modest body of research has been conducted to investigate public safety communications. Stein et al. focus on resource management using ASR in crisis management and emergency responses, suggesting a great potential for this application [41]. Additionally, Stein et al. (in a separate work the same year) investigated ASR and keyword extraction with firefighter radio communications, reinforcing the perceived potential for applications in this sector [42]. As part of the same study, it was noted that there were potential difficulties with the complex, coded communications and variable quality of broadcasts, a topic further investigated by Bochner et al. [43]. Srivastava et al. performed a similar study with police radio communications in the greater Chicago area, suggesting that out-of-the-box models are good at transcribing coded communications, but can be improved with further tuning [44].
In the context of traffic incident management (TIM), there is a notable lack of existing literature pertinent to public safety communications. Though proposed by Zheng et al. in a speculative work regarding applications of large language models (LLMs) in transportation [45], no defining works or studies have been published for this specific use.

2.6. Study Location

This study was conducted using a digital police scanner located at Purdue University’s West Lafayette, Indiana campus. The site “listened” to (monitored) 9-1-1 dispatch centers from four counties nearby: White, Tippecanoe, Clinton, and Boone, shown in Figure 1 below. The figure illustrates the alignment of I-65 (red line), the approximate linear reference for county border crossings (blue halo) at MMs 128, 150, 160, 184, and 199, and civil townships (green halo). The defined study area consists of 71 centerline miles of Interstate I-65 (red), a primarily rural Interstate connecting Chicago, Illinois, and Indianapolis, Indiana.

3. Description of Incident Used for Case Study

A case study is presented to demonstrate the vital information that 9-1-1 dispatches contain. Figure 2 depicts a tractor-trailer fire occurring on 27 June 2025, on I-65 northbound near the 165-mile marker (MM). The fire-damaged vehicle, first responders, and their equipment are denoted by callouts i, ii, and iii, respectively. This incident resulted in a full closure of both directions of travel and an estimated 10-mile queue in the northbound travel lanes. Despite creating a 10-mile queue, this incident was never reported in the state Motorist Information System, most likely due to its rural location.

9-1-1 Dispatch Audio and Transcribed Text for Truck Fire Incident

Table 1 presents the captured 9-1-1 dispatch and operations communications relevant to the incident introduced in Figure 2. The hyperlink in the caption of Table 1 is the audio recording of the dispatch of events. The initial dispatch was captured at 14:59:14 local time and transcribed with a 2% WER. This transmission captures “who” (Sheffield Fire (Tippecanoe County)), “what” (a semi-fire), “when” (14:59), and “where (I-65 northbound from the 165 MM). This transmission conveys all the essential information necessary for a TMC operator to be informed and begin monitoring for travel impacts that require motorist alerts through the MIS.
While the initial transmission provides abundant information, the subsequent operations conversations provide further information, for example:
  • Location confirmation at 15:09 to be at a specific exit.
  • The closure of travel lanes at 15:13.
  • The opening of travel lanes at 15:28.
  • The scene cleared at 15:45.

4. Methodology

The methodology section discusses the proposed methodologies for audio capture, audio transcription, keyword extraction, and automated email notification. The authors employed a combination of an open-access transcription model, a commercial radio scanner, and common enterprise tools to construct the prototype.

4.1. System Integration Architecture

Figure 3 is the generalized architecture used to generate automated email notifications. It is divided into four main “segments.”
I.
Audio transmission.
II.
Audio capture.
III.
Audio transcription and keyword extraction.
IV.
Notification and action.
Subfigure (a) represents the independently broadcasted messages from local 9-1-1 dispatch centers; subfigures (b,c) represent the radio capture processes using commercial equipment and software; (d,e) represent open access transcription software combined with business process software to generate automated emails and subfigures (f,g) represent traffic management activities after a notification is generated by separate personnel. The remaining subsections of this section will discuss subfigures (b–e) in more detail.

4.2. Capturing Audio

The Uniden SDS200 Digital Police Scanner (BearcatWarehouse, Uniden, Elkridge, MD, USA) (Figure 4) was used to capture audio, scanning frequencies in the 800 MHz band. The SDS200 is an industry-standard unit used for scanning and capturing public safety communications in the United States. The unit has an effective range of about 30 miles, subject to several factors [46]. The unit was used in combination with the “ProScan All-In-One Computer Aided Scanning Program” [47]. ProScan facilitated the recording, indexing, and programming of the SDS200 unit. Information passed between the SDS200 and ProScan software includes (not exhaustively) the system (callout i), the department and channel (callout ii), and the exact frequency (callout iii). Data and audio are transferred over LAN using the connection shown as callout iv.
Targeted channels were selected to provide the most likely opportunity to capture a 9-1-1 dispatch to an interstate incident. Selected channels are presented below in Table 2 with the County/Radio system to which they belong; the department, channel, and type; the approximate MM; and whether the system is conventional (conv.) or trunked (trun.). With the addition of 9-1-1 dispatch channels, a select group of operations (ops) channels was included, as well as the state police for the region. Since this area of the state still utilizes a combination of legacy conventional fixed frequencies and modern digital trunked systems, it was essential to monitor both to provide comprehensive coverage. Once the appropriate systems, departments, and channels were identified, the Uniden Sentinel Software could be used to program the digital scanner.

4.3. Transcribing Audio

Once captured, the audio must be transcribed. To perform these transcriptions, the Python package “openai-whisper” (version 20240930) [48], specifically, the “turbo” model (an optimized version of the “large-v3” model) was used. Files were synced from the remote scanning PC to a central PC using Dropbox, although any means of sharing can be utilized (or transcription performed on the recording PC). Although OpenAI-Whisper was used for this study, there are a variety of excellent open-source and pay-per-use transcription models that could also have been used with similar expected performance.

4.4. Keyword Extraction and Notification

From the transcribed audio (a string of text), Regular Expressions (RegEx) are used in combination with simple logic to extract instances of keywords belonging to one of three groups, as shown in Table 3 below.
  • “Interstate” keywords refer to specific references to an interstate. For the study location, the only interstate of interest is I-65; hence, the examples are limited to I-65 (including the local phrase “the I”). “Interstate” keywords serve as the primary indicator of an incident occurring on the interstate.
  • “Location” keywords refer to a particular location along the interstate. The desired “location” is the MM (e.g., ‘168.5’); however, sufficient information is contained within specific exits/entrances to infer most locations along a given route. Additional context, including which lane or shoulder, is also considered valuable as it will expedite the locating time for a TMC operator.
  • “Incident” keywords refer to additional context about the specific incident (e.g., ‘fire’ or ‘crash’). These keywords help indicate the type and severity of the incident.
In addition to the common phrases presented in the table, coded communications are also included. Commonly referred to as “ten-codes,” these phrases convey information such as units on scene, scenes being cleared, or other frequently used messages.
A simple logical structure is used to determine if the keywords present in any given audio message warrant a notification to relevant parties. Figure 5 depicts this structure.
Starting at the new audio section, transcriptions are checked for an “interstate” reference to determine if it pertains to an interstate or not. Any additional context is then pulled from the “incident” keywords. Finally, “location” indicators are extracted, with preference given to mile marker utterances. For a message to be considered “important enough” to be sent (Figure 3e), it must contain an “interstate” reference and an “incident” reference; “location” indicators are not strictly required as per this criterion.
Transcriptions further down the flow diagram in Figure 5 have “more context,” and thus contain more pertinent information when delivered to a TMC operator. The ideal transmission, based on this structure, specifically mentions an interstate, incident type, and the exact MM.

5. Results

The following section presents the study’s results. First, all incidents detected in the study area (Figure 1) are presented, followed by a detailed case study (as introduced in Figure 2).
This study was conducted over a period of two months, from 12 May to 30 June, 2025. In total, over 100,000 audio transmissions were captured across the four counties in Figure 1. Figure 6 summarizes the incidents identified having an “interstate” keyword with (or without) a “location” keyword. Callout i refers to the incident detailed in the case study that follows. Callout ii denotes the period prior to the start of data collection. From the more than 100,000 transmissions, 87 transmissions were captured and indicated with some variation in “interstate” and other keyword groups from Table 3.
Of the 87 identified transmissions, 37 were assigned a geospatial position using the mile marker. Another 40 were manually assigned a geospatial position (where the majority were not automatically placed due to being adjacent to the interstate), and 10 were unable to be located. These 76 identified transmissions are plotted in Figure 7 below, where the incident referenced in Figure 6, callout i, is given by callout i. This result suggests that the quality of transcriptions and the RegEx extraction provides a feasible solution for monitoring 9-1-1 dispatch channels and automatically extracting relevant information accurately.

6. Case Study & Discussion

This section will further detail the incident introduced in Figure 1 and Figure 2. The incident location is shown as callout i in Figure 7.

6.1. Automated Notification

Figure 8 is the automated email notification corresponding to the incident in this case study. It was generated entirely autonomously using Microsoft Power Automate, Microsoft SharePoint, Python, and the recording devices and software described in the Capturing Audio Section and corresponds to the incident described in Figure 2.
Figure 8 is an example of an ideal transcription. Callout iii is the transcription generated from the ASR engine. Blue words indicate identified keyword indicators. In the observed audio clips, repetition of information is common in dispatch transmissions. Thus, Figure 5 is used to identify the “best” indicators and present them in the section denoted by callout iv. Finally, the recording contains useful information within its metadata. Callout v is data captured during the recording process, including the date and time of capture, as well as the site, department, and transmission channel. Information from callouts iv and v allows for the localization of transmissions to a specific geographic area, and in ideal cases, a particular location along an interstate.
The median latency between the time an audio file was recorded (callout i) and the time the email was received (callout ii) was approximately 52 s. This suggests that the proposed methodology acts quickly enough to relay important information before it becomes outdated.

6.2. Spatiotemporal Heatmaps

Spatiotemporal traffic speed plots (heatmaps) present linearly referenced traffic speed trajectories, color-coded by speed [49]. Heatmaps are important tools that can be used for both real-time monitoring of freeway conditions and detailed “after-action” analysis of roadway incidents. Figure 9 is the heatmap corresponding to the incident in Figure 2.
Figure 9a presents the heatmaps for both northbound and southbound directions. Callout i indicates the direction of travel; callout ii indicates the period of interest corresponding to the case study. Prior to the incident, a storm affected travel behavior on the same segment as the incident. This is observed as ‘patches’ on the heatmap and is noted by callout iii for both directions of travel. Callout iv indicates reduced speeds and queuing due to the incident blocking traffic in the northbound lanes. Callout v indicates reduced speeds and queuing due to the incident response in the southbound lanes. Figure 9b isolates the primary incident in the northbound lanes. From the heatmap, the following observations are made about the incident, and verified through the events of Table 1:
  • The incident occurred at MM 165.8, northbound (callout vii).
  • The longest queue reached MM 156.1, an estimated 9.6-mile queue (callout vi).
  • The incident began at 14:59 (callout vii).
  • There is a short-term NB closure at approximately 15:22 near MM 165.8 (callout ix). The transcribed text for this can be seen in Table 1 at 15:22:08, and the lack of vehicles passing through this area is shown as the white area on the northbound heatmap at this approximate time.
  • The incident cleared at 15:45 (callout viii).
  • There were two ITS cameras near the incident (callout x and callout xi).
  • The queue reached the upstream camera (callout xi) and cleared the same, as shown by callout xii and xiii, respectively.
  • ITS images from those cameras are shown in Figure 10 and are described in more detail below.

6.3. ITS Camera Images

Figure 10 shows the images from two ITS cameras, downstream (Figure 9, callout x) and upstream (Figure 9, callout xi) of the incident. Figure 10 approximates the information a TMC operator would have to investigate the occurrence of an incident without 9-1-1 dispatch audio or phone calls. For this case study, the incident occurred outside the viewable area for both cameras. Notable events from the subfigures of Figure 10 are as follows:
(a)
Incident occurrence, see the capture time (callout i).
(b)
A fire engine entered the southbound lanes (callout ii, being the same as Table 1, row “15:08:20”). Vehicles are traveling in the northbound lanes (callout iii) at nine minutes after the incident starts.
(c)
Full closure of the northbound lanes at twenty minutes after the incident starts.
(d)
Northbound lanes reopened thirty-one minutes after the incident started (callout iv).
(e)
Upstream camera captures the queue exceeding its location (callout v) at fifty-seven minutes after the incident occurs.
(f)
The queue clears the upstream camera (callout vi) at eighty-four minutes after the incident starts.

6.4. Enhancing TMC Incident Awareness of Remote Incidents with Dispatch Audio

While the ITS camera images of Figure 10 provide some information on the queuing. The full lane closures of the northbound lanes and southbound lanes were not visible. Recorded 9-1-1 dispatch and operations communications (Table 1) provide a detailed chronology (even when not complete), unmatched by the heatmap (Figure 9) or ITS images (Figure 10).
The results of the study suggest that the ASR of public safety dispatch centers, particularly in remote areas, can extend the visibility that TMCs have on rural sections of interstates. In fact, ASR is a natural extension of what TMC operators already do when monitoring dispatch channels in metropolitan areas.

7. Conclusions

This paper presents the findings from a pilot study conducted over two months, spanning four counties, aimed at assessing the feasibility of using off-the-shelf equipment and Automatic Speech Recognition models to provide timely coverage of rural interstate incidents to TMC operators.
This paper proposed the use of transcribing local 9-1-1 dispatch and operations communications captured over public safety radio broadcasts from a single centrally located scanning site, shown in Figure 1. Using unmodified, commercial, and open-source technology, 71 miles of I-65 are monitored 24/7, providing near-real-time coverage of incidents as 9-1-1 is notified. During the study period (May–June 2025), over 100,000 transmissions were captured and transcribed, generating 76 locatable events. Although the word error rate (WER) shown in Table 1 showed some transcription errors, the transmissions generally provided rich insight into the time, location, severity, and ultimately duration of an incident. Figure 8, Figure 9 and Figure 10 provided both quantitative validation of the incident time, location, severity, and duration, as well as images from ITS cameras that showed qualitatively the entrance of emergency vehicles and the extent of queuing.
The findings of this study indicate that modern Automatic Speech Recognition models, cloud storage solutions, commercial computers, scanners, and software can easily be combined to develop a system capable of monitoring dozens of miles of rural interstate, providing near-real-time incident notifications 24/7 to TMC operators. Furthermore, additional research is warranted to further advance the proposed system.

Future Opportunities

Although this demonstrates the feasibility of covering a 71-mile section of Interstate, further work is needed to develop a well-packaged system that can be quickly deployed at several sites adjacent to critical corridors and integrated into a cloud-based system. Such integration will involve further optimization of audio transcription models, perhaps with specialized training on “ten” codes and “signal” codes commonly used in public safety communication. This could also facilitate integration with large-language models (LLMs) in the management of knowledge and assessment of transmissions. Building out such an infrastructure would enable traffic management centers to cost-effectively extend their coverage by strategically leveraging emerging artificial intelligence to augment existing traffic management staff, allowing them to focus on the most critical incidents without the challenges of monitoring routine audio transmissions.
Before scalable systems, further research is warranted to validate the transcription accuracy, the incident detection reliability (i.e., are any broadcasts missed?), the geolocation, precision, and accuracy. This study does not provide any quantitative metrics to compare methodologies with each other or assess their performance in terms of what is transmitted. This course of study should be first addressed before any of the scalability issues.

Author Contributions

The authors confirm contribution to the paper as follows: study conception and design: C.M.G., V.V. and D.M.B.; data collection: C.M.G., V.V., J.D. and D.M.B.; analysis and interpretation of results: C.M.G., V.V. and D.M.B.; draft manuscript preparation: C.M.G., V.V., J.D. and D.M.B. All authors have read and agreed to the published version of the manuscript.

Funding

This study is based upon work supported by the Joint Transportation Research Program administered by the Indiana Department of Transportation and Purdue University.

Acknowledgments

The Uniden SDS200 Digital Police Scanner was used to monitor the public safety audio. The Pro-Scan Software v23.4 was used to digitally record the audio. The OpenAI Whisper ASR engine, specifically the “Turbo, large-v3” model (implemented using the openai-whisper Python package), was utilized to provide speech-to-text transcription of the audio recorded by the Pro-Scan Software. The authors wish to acknowledge the efforts of George H. Goble, Howell Li, Rahul Suryakant Sakhare, and Daniel Saldivar-Carranza in the progress of this research. This work was supported by the Joint Transportation Research Program and the Indiana Department of Transportation. The contents of this paper reflect the views of the authors, who are responsible for the facts and the accuracy of the data presented herein, and do not necessarily reflect the official views or policies of the sponsoring organization. The authors affirm that no AI or LLMs were used in any capacity in the drafting of the contents of this manuscript.

Conflicts of Interest

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

References

  1. INDOT Condition Acquisition and Reporting System (CARS) & TMC Reporting Policy 2021. Available online: https://www.in.gov/indot/files/INDOT_CARS_Policy.pdf (accessed on 1 July 2025).
  2. 511 Deployment Coalition (U.S.). America’s Travel Information Number: Implementation and Operational Guidelines for 511 Services 2005; 511 Deployment Coalition (U.S.): Washington, DC, USA, 2005. [Google Scholar]
  3. Burgess, L.M.; Garinger, A.; Carrick, G. Integrating Computer-Aided Dispatch Data with Traffic Management Centers; Cambridge Systematics, Inc.: Medford, MA, USA, 2021. [Google Scholar]
  4. Judycki, D.C.; Robinson, J.R. Managing Traffic during Nonrecurring Congestion. ITE J. 1992, 1992, 21–26. [Google Scholar]
  5. Transportation Research Board. Better Use of Existing Transportation Facilities. In Proceedings of the Seventh Summer Meeting of the Transportation Research Board in Cooperations with the Florida Department of Transportation, Jacksonville, FL, USA, 5–7 August 1974; Transportation Research Board National Research Council: Jacksonville, FL, USA, 1975.
  6. Cook, A.R.; Cleveland, D.E. The Detection of Freewauy Capactiy Reducing Incidents by Traffic Stream Measurements. In Optimizing Freeway Corridor Operations Through Traffic Surveillance, Communication and Control; The University of Michigan, Highway Safety Research Institute: Ann Arbor, MI, USA, 1970. [Google Scholar]
  7. Payne, H.J. Freeway iIncident Detection Based Upon Pattern Classification. In Proceedings of the 1975 IEEE Conference on Decision and Control including the 14th Symposium on Adaptive Processes, Houston, TX, USA, 10–12 December 1975; pp. 688–692. [Google Scholar]
  8. Willsky, A.; Chow, E.; Gershwin, S.; Greene, C.; Houpt, P.; Kurkjian, A. Dynamic Model-Based Techniques for the Detection of Incidents on Freeways. IEEE Trans. Autom. Control 1980, 25, 347–360. [Google Scholar] [CrossRef]
  9. Cheu, R.L.; Ritchie, S.G. Automated Detection of Lane-Blocking Freeway Incidents Using Artificial Neural Networks. Transp. Res. Part C Emerg. Technol. 1995, 3, 371–388. [Google Scholar] [CrossRef]
  10. Gangisetty, R. Advanced Traffic Management System on I-476 in Pennsylvania. In Proceedings of the Conference on Intelligent Transportation Systems, Boston, MA, USA, 12 November 1997; pp. 373–378. [Google Scholar]
  11. Champion, H.R.; Augenstein, J.; Blatt, A.J.; Cushing, B.; Digges, K.; Siegel, J.H.; Flanigan, M.C. Automatic Crash Notification and the URGENCY Algorithm. Adv. Emerg. Nurs. J. 2004, 26, 143–156. [Google Scholar]
  12. Oh, J.-S.; Oh, C.; Ritchie, S.G.; Chang, M. Real-Time Estimation of Accident Likelihood for Safety Enhancement. J. Transp. Eng. 2005, 131, 358–363. [Google Scholar] [CrossRef]
  13. Alexander, L.; Cheng, P.-M.; Gorjestani, A.; Menon, A.; Newstrom, B.; Shankwitz, C.; Donath, M. The Minnesota Mobile Intersection Surveillance System. In Proceedings of the 2006 IEEE Intelligent Transportation Systems Conference, Toronto, ON, Canada, 17–20 September 2006; pp. 139–144. [Google Scholar]
  14. Oh, C.; Park, S.; Ritchie, S.G. A Method for Identifying Rear-End Collision Risks Using Inductive Loop Detectors. Accid. Anal. Prev. 2006, 38, 295–301. [Google Scholar] [CrossRef] [PubMed]
  15. Thompson, C.; White, J.; Dougherty, B.; Albright, A.; Schmidt, D.C. Using Smartphones to Detect Car Accidents and Provide Situational Awareness to Emergency Responders. In Mobile Wireless Middleware, Operating Systems, and Applications; Cai, Y., Magedanz, T., Li, M., Xia, J., Giannelli, C., Eds.; Springer: Berlin/Heidelberg, Germany, 2010; pp. 29–42. [Google Scholar]
  16. Gongjun, Y.; Olariu, S.; Popescu, D.C. NOTICE: An Architecture for the Notification of Traffic Incidents. IEEE Intell. Transp. Syst. Mag. 2012, 4, 6–16. [Google Scholar] [CrossRef]
  17. Junior, J.G.R.; Campista, M.E.M.; Costa, L.H.M.K. A Decentralized Traffic Monitoring System Based on Vehicle-to-Infrastructure Communications. In 2013 IFIP Wireless Days (WD); IEEE: Valencia, Spain, 2013; pp. 1–6. [Google Scholar]
  18. Foggia, P.; Saggese, A.; Strisciuglio, N.; Vento, M.; Petkov, N. Car Crashes Detection by Audio Analysis in Crowded Roads. In Proceedings of the 2015 12th IEEE International Conference on Advanced Video and Signal Based Surveillance (AVSS), Karlsruhe, Germany, 25–28 August 2015; pp. 1–6. [Google Scholar]
  19. Almaadeed, N.; Asim, M.; Al-Maadeed, S.; Bouridane, A.; Beghdadi, A. Automatic Detection and Classification of Audio Events for Road Surveillance Applications. Sensors 2018, 18, 1858. [Google Scholar] [CrossRef] [PubMed]
  20. Apple Inc. Use Crash Detection on iPhone or Apple Watch to Call for Help in an Accident. Available online: https://support.apple.com/en-us/104959 (accessed on 17 July 2025).
  21. Senarath, Y.; Nannapaneni, S.; Purohit, H.; Dubey, A. Emergency Incident Detection from Crowdsourced Waze Data Using Bayesian Information Fusion. In Proceedings of the 2020 IEEE/WIC/ACM International Joint Conference on Web Intelligence and Intelligent Agent Technology (WI-IAT), Melbourne, Australia, 14–17 December 2020; pp. 187–194. [Google Scholar]
  22. Pan, D.; Hamdar, S. From Traffic Analysis to Real-Time Management: A Hazard-Based Modeling for Incident Durations Extracted Through Traffic Detector Data Anomaly Detection. Transp. Res. Rec. 2024, 2678, 389–400. [Google Scholar] [CrossRef]
  23. Xiao, D.; Wang, Z.; Shen, Z.; Xu, X.; Ma, C. A FairMOT Approach Based on Video Recognition for Real-Time Automatic Incident Detection on Expressways. Signal Image Video Process. 2024, 18, 7333–7348. [Google Scholar] [CrossRef]
  24. ElSahly, O.; Abdelfatah, A. Developing a Machine-Learning-Based Automatic Incident Detection System for Traffic Safety: Promises and Limitations. Infrastructures 2024, 9, 170. [Google Scholar] [CrossRef]
  25. Zdunek, K. Design Considerations of Trunked Radio Systems. In Proceedings of the 36th IEEE Vehicular Technology Conference, Dallas, TX, USA, 20–22 May 1986; Volume 36, pp. 194–201. [Google Scholar]
  26. Jones, L.T. Trunked Radio Systems; U.S. Department of Justice, Office of Justice Programs, National Institute of Justice, Technology Assessment Program: Washington, DC, USA, 1993. [Google Scholar]
  27. Trunked Radio Systems. Available online: https://wiki.radioreference.com/index.php/Trunking (accessed on 9 July 2025).
  28. Sharp, D.S.; Cackov, N.; Laskovic, N.; Shao, Q.; Trajkovic, L. Analysis of Public Safety Traffic on Trunked Land Mobile Radio Systems. IEEE J. Sel. Areas Commun. 2004, 22, 1197–1205. [Google Scholar] [CrossRef]
  29. What Is Project Hoosier SAFE-T? Available online: https://faqs.in.gov/hc/en-us/articles/115005056187-What-is-Project-Hoosier-SAFE-T (accessed on 27 May 2025).
  30. What Evidence Is Available Showing That the SAFE-T Network Works? Available online: https://faqs.in.gov/hc/en-us/articles/115005228848-What-evidence-is-available-showing-that-the-SAFE-T-network-works (accessed on 9 July 2025).
  31. About RadioReference. Available online: https://www.radioreference.com/about/ (accessed on 9 July 2025).
  32. Tippecanoe County (IN)—The RadioReference Wiki. Available online: https://wiki.radioreference.com/index.php/Tippecanoe_County_(IN) (accessed on 23 June 2025).
  33. TETRA. Available online: https://www.etsi.org/technologies/tetra (accessed on 17 September 2025).
  34. IBM100—Pioneering Speech Recognition. Available online: https://web.archive.org/web/20150219080748/http://www-03.ibm.com/ibm/history/ibm100/us/en/icons/speechreco/ (accessed on 9 July 2025).
  35. Huang, X.; Baker, J.; Reddy, R. A Historical Perspective of Speech Recognition. Commun. ACM 2014, 57, 94–103. [Google Scholar] [CrossRef]
  36. Ibrahim, H.; Varol, A. A Study on Automatic Speech Recognition Systems. In Proceedings of the 2020 8th International Symposium on Digital Forensics and Security (ISDFS), Beirut, Lebanon, 1–2 June 2020; pp. 1–5. [Google Scholar]
  37. Popović, M.; Ney, H. Word Error Rates: Decomposition over Pos Classes and Applications for Error Analysis. In Proceedings of the Second Workshop on Statistical Machine Translation—StatMT ’07, Prague, Czech Republic, 23 June 2007; Association for Computational Linguistics: Prague, Czech Republic, 2007; pp. 48–55. [Google Scholar]
  38. Wang, Y.-Y.; Acero, A.; Chelba, C. Is Word Error Rate a Good Indicator for Spoken Language Understanding Accuracy. In Proceedings of the 2003 IEEE Workshop on Automatic Speech Recognition and Understanding, St. Thomas, VI, USA, 30 November–4 December 2003; pp. 577–582. [Google Scholar]
  39. Stehr, R.A. Minnesota Department of Transportation Experience in the Application of Advanced Traffic Management Systems; ASCE: Minneapolis, MN, USA, 1991; pp. 458–462. [Google Scholar]
  40. Sakhare, R.S.; Desai, J.; Woker, W.; Li, H.; Mathew, J.K.; Mahlberg, J.; Saldivar-Carranza, E.D.; Horton, D.; Bullock, D.M.; United States Department of Transportation. Intelligent Transportation Systems Joint Program Office; et al. Connected Vehicle-Centric Dashboards for TMC of the Future; SPR-4396; Purdue Universityl: West Lafayette, IN, USA, 2023; p. 41. [Google Scholar]
  41. Stein, D.; Krausz, B.; Löffler, J.; Marterer, R.; Bardeli, R.; Schwenninger, J.; Usabaev, B. Enriching an Intelligent Resource Management System with Automatic Event Recognition. In Proceedings of the 9th International ISCRAM Conference, Vancouver, BC, Canada, 22–25 April 2012. [Google Scholar]
  42. Stein, D.; Schwenninger, J.; Usabaev, B. Automatic Chat Transcription on a Firefighter TETRA Broadcast Channel. In Proceedings of the Speech Communication, 10. ITG Symposium, Braunschweig, Germany, 26–28 September 2012; pp. 1–4. [Google Scholar]
  43. Bochner, J.; Indelicato, M.; Konnur, P. Effects of Sound Quality on the Accuracy of Telephone Captions Produced by Automatic Speech Recognition: A Preliminary Investigation. Am. J. Audiol. 2023, 32, 243–250. [Google Scholar] [CrossRef] [PubMed]
  44. Srivastava, T.; Chou, J.-C.; Shroff, P.; Livescu, K.; Graziul, C. Speech Recognition for Analysis of Police Radio Communication. In Proceedings of the 2024 IEEE Spoken Language Technology Workshop (SLT), Macao, China, 2–5 December 2024; pp. 906–912. [Google Scholar]
  45. Zheng, O.; Abdel-Aty, M.; Wang, D.; Wang, Z.; Ding, S. ChatGPT Is on the Horizon: Could a Large Language Model Be All We Need for Intelligent Transportation? arXiv 2023, arXiv:2303.05382. [Google Scholar]
  46. SDS200 Owner’s Manual. Available online: https://www.uniden.info/download/ompdf/SDS200om.pdf (accessed on 9 July 2025).
  47. Bob Aune ProScan. Available online: https://www.proscan.org/ (accessed on 9 July 2025).
  48. Openai/Whisper. Available online: https://github.com/openai/whisper (accessed on 9 July 2025).
  49. Sakhare, R.S.; Desai, J.; Mathew, J.K.; McGregor, J.; Kachler, M.; Bullock, D.M. Measuring and Visualizing Freeway Traffic Conditions: Using Connected Vehicle Data; Joint Transportation Research Program: West Lafayette, IN, USA, 2024. [Google Scholar]
Smartcities 08 00157 g001
Figure 1. Study location spanning seventy-one miles of I-65, capturing four county dispatch regions.
Figure 1. Study location spanning seventy-one miles of I-65, capturing four county dispatch regions.
Smartcities 08 00157 g001
Smartcities 08 00157 g002
Figure 2. First responders work a rural interstate trailer fire incident (Captured by Kameron Pelfree, 27 June 2025).
Figure 2. First responders work a rural interstate trailer fire incident (Captured by Kameron Pelfree, 27 June 2025).
Smartcities 08 00157 g002
Smartcities 08 00157 g003
Figure 3. Flow diagram for public safety radio communication recording, archiving, transcribing, contextualizing, and alerting: (a) 9-1-1 dispatch; (b) audio capture; (c) audio storage; (d) audio transcription and keyword extraction; (e) notification generation; (f) further action; (g) MIS alert.
Figure 3. Flow diagram for public safety radio communication recording, archiving, transcribing, contextualizing, and alerting: (a) 9-1-1 dispatch; (b) audio capture; (c) audio storage; (d) audio transcription and keyword extraction; (e) notification generation; (f) further action; (g) MIS alert.
Smartcities 08 00157 g003
Smartcities 08 00157 g004
Figure 4. The Uniden SDS200 digital police scanner is used to monitor radio traffic.
Figure 4. The Uniden SDS200 digital police scanner is used to monitor radio traffic.
Smartcities 08 00157 g004
Smartcities 08 00157 g005
Figure 5. Flow diagram for keyword extraction and contextualization of transcribed audio.
Figure 5. Flow diagram for keyword extraction and contextualization of transcribed audio.
Smartcities 08 00157 g005
Smartcities 08 00157 g006
Figure 6. Stacked bar plot of keyword-flagged transcriptions with an “incident” grouped by other keywords detected for all data collected in May and June 2025.
Figure 6. Stacked bar plot of keyword-flagged transcriptions with an “incident” grouped by other keywords detected for all data collected in May and June 2025.
Smartcities 08 00157 g006
Smartcities 08 00157 g007
Figure 7. Location map of captured incidents for all data collected in May and June 2025.
Figure 7. Location map of captured incidents for all data collected in May and June 2025.
Smartcities 08 00157 g007
Smartcities 08 00157 g008
Figure 8. Sample automated notification generated corresponding to the trailer fire in Figure 2.
Figure 8. Sample automated notification generated corresponding to the trailer fire in Figure 2.
Smartcities 08 00157 g008
Smartcities 08 00157 g009
Figure 9. Spatiotemporal traffic speed plot (“heatmap”) of I-65 MM 150–180, focused on the impacts of the event presented in Figure 2. (a) heatmap identifying incident and bidirectional effects. (b) heatmap limited to primary incident (red bounding box).
Figure 9. Spatiotemporal traffic speed plot (“heatmap”) of I-65 MM 150–180, focused on the impacts of the event presented in Figure 2. (a) heatmap identifying incident and bidirectional effects. (b) heatmap limited to primary incident (red bounding box).
Smartcities 08 00157 g009
Smartcities 08 00157 g010aSmartcities 08 00157 g010b
Figure 10. ITS camera images of significant events in chronological order for the trailer fire incident. Images bounded in yellow correspond to the “downstream camera” and those in blue correspond to the “upstream camera“. (a) Incident occurs at 14:58:27 (T-1 min). (b) Fire engine(s) seen entering the interstate through the southbound entrance ramp at 15:08:27 (T + 9 min). (c) Full closure of directions of travel at 15:18:28 (T + 20 min). (d) Partially opening in both directions of travel at 15:30:27 (T + 31 min). (e) Northbound back-of-queue reaches the upstream camera at 15:56:19 (T + 57 min); (f) Northbound back-of-queue clears the upstream camera at 16:24:20 (T + 84 min).
Figure 10. ITS camera images of significant events in chronological order for the trailer fire incident. Images bounded in yellow correspond to the “downstream camera” and those in blue correspond to the “upstream camera“. (a) Incident occurs at 14:58:27 (T-1 min). (b) Fire engine(s) seen entering the interstate through the southbound entrance ramp at 15:08:27 (T + 9 min). (c) Full closure of directions of travel at 15:18:28 (T + 20 min). (d) Partially opening in both directions of travel at 15:30:27 (T + 31 min). (e) Northbound back-of-queue reaches the upstream camera at 15:56:19 (T + 57 min); (f) Northbound back-of-queue clears the upstream camera at 16:24:20 (T + 84 min).
Smartcities 08 00157 g010aSmartcities 08 00157 g010b
Table 1. Transcription of captured audio from an I-65 truck fire occurring on 27 June 2025 near mile marker 165 in Indiana [https://youtu.be/SJrGKEPGSg4] (accessed on 23 Jul 2025).
Table 1. Transcription of captured audio from an I-65 truck fire occurring on 27 June 2025 near mile marker 165 in Indiana [https://youtu.be/SJrGKEPGSg4] (accessed on 23 Jul 2025).
TimeChannelTranscriptionActionWER
14:59:14Co. DispatchAttention Sheffield fire. Sheffield fire. Have a semi whose rear axle is on fire. Northbound I-65 from the 165 mile marker. Allegedly still driving northbound. Callers are trying to get him to pull over currently. Again northbound from the 165. It’s a semi-pulling [a] box trailer. Rear axle on the driver’s side is on fire. Tone time is 1459.Dispatch broadcasted
[Automated Notification Trigger]		0.02
15:06:12Co. Dispatch204 [en] route [with] 2. 204.Units en route0.5
15:08:08Sheffield FD204 on Sheffield. 0.0
15:08:20Sheffield FDTo the [interstate] until we find out where we’re going. 0.0
15:09:40Co. DispatchI’m on the phone with the semi-driver now. He’s pulled over. Looks like near the [Wyandot] overpass. [Advising] tandems are still on fire.Location Confirmed0.09
15:10:05Co. Dispatch204 is clear [Wyandot] overpass. 206 go ahead and make access to the 65 now. 0.13
15:11:14Co. Sheriff’s Office23. Go ahead. Do you know where the semi-driver’s [at]? Traffic’s already backed up to the 162. He’s right at the [Wyandot] overpass. Sorry.Driver Located0.13
15:13:20Co. DispatchContainer fire 0.50
15:13:40Co. DispatchTippe. Fire 220. 204 is arriving on scene working fire. 220 will have IC on 65. Northbound they’re going to be shut down by ISP. []Units on Scene0.08
15:14:53Sheffield FD220 40 on Sheffield. We have three on station. What do you want? 0.00
15:16:11Co. Dispatch206 where do you want me to stage? Stay high [and] forward. Clear. 0.08
15:16:29Sheffield FD2040 on Sheffield. You want to throw a one or a tanker? 0.00
15:16:43Co. DispatchWe got a good knockdown on it.Fire out0.13
15:22:08Co. DispatchIn the northbound lanes we’ve got it completely shut down for now.NB closure0.00
15:22:26Sheffield FDBe southbound and northbound.SB closure0.00
15:23:42Co. Dispatch One when you arrive just go ahead and get turned around and face back north in front of the. Come on. 0.00
15:28:08Co. SOWe can open up the left lane. Sure.Reopening [NB] lane0.00
15:28:39Co. DispatchGo ahead. Are they able to open up both lanes or just northbound or southbound? You’ve got a definite block in northbound. He can let the high-speed side go. We’ll keep the low speeds shut down still. 0.00
15:28:58Co. DispatchClear. Further southbound slowing is just slow for some reason. Been like that for almost an hour now.Confirmation SB open0.00
15:29:16Co. SOGot it. We’ll be clear. Clear. 0.00
15:45:03Co. DispatchTippe. fire 220. 220. Command [survey] on I-65, all [seems to] be cleared in the scene and service. I’ll move the pressure. Thank you.Units clear scene0.09
Table 2. Summary of radio systems, departments, and channels and their approximate linear reference coverage by county and department for conventional and trunked systems used.
Table 2. Summary of radio systems, departments, and channels and their approximate linear reference coverage by county and department for conventional and trunked systems used.
Sys. DepartmentChannelTypeMMConvTrun
SAFE-TState Police Region 1—LowellISP D14 Lafayette Disp.P🚨150–199 ✓
ISP D14 Lafayette MultigroupP🚨 ✓
ISP D14 Lafayette Ops. [1,2,3]P🚨 ✓
SAFE-T.White Co.Co. Fire Disp.F🔥184–199✓✓
Co. (FD) Ops. [1,2,3,4,5,6,7]F🔥 ✓
Sherrif Disp.P🚨 ✓
Co. EMS Ops. 2E🚑 ✓
Tippecanoe Co.Tippecanoe Co. Emergency Ambulance ServiceEMS Disp. and Alt.E🚑160–184✓✓
Tippecanoe Co. Emergency Mngmt.Disp.G🏛️ ✓
Tippecanoe Co. S.O.Disp.P🚨 ✓
Tippecanoe Co. Emergency Mngmt./Fire/EMSFire Disp./Tone-OutsF🔥✓
Tippecanoe Co. FireDisp.F🔥 ✓
Fire Mutual Response AreaF🔥
Battleground FDF🔥176–184 ✓
Lafayette FDSta. 6 Ops.F🔥174–176 ✓
Sta. 5 Ops.F🔥172–174 ✓
Sta. 9 Ops.F🔥170–172 ✓
Sta. [2,7] Ops.F🔥* ✓
Tippecanoe Co. FireSheffield FDF🔥160–170 ✓
Clark Hill FDF🔥* ✓
SAFE-TClinton Co.Co. Fire Disp.F🔥150–160✓✓
Co. Law Enforcement Disp. [1,2]P🚨✓✓
SAFE-TBoone Co.Co. EMA/EMS Ops.E🚑128–150✓✓
Co. Fire Disp.F🔥✓✓
Co. Fire Ops. [1,2,3,4,5]F🔥 ✓
Sherrif Disp. [1,2]P🚨 ✓
(F)🔥: Fire; (P)🚨: Law Enforcement; (E)🚑: EMS; (G)🏛️: Government/Other* Mutual Aide Only
Table 3. Keyword extraction groups, examples, and descriptions.
Table 3. Keyword extraction groups, examples, and descriptions.
Keyword GroupExamplesDescription
“Interstate”“interstate”, “the i”, “i-65”, “65”Common and local phrase extraction for Interstate 65 (I-65) running through the study area.
“Location”“168.5 northbound”, “155 mile marker”, “exit”, “left shoulder”The locational information contained within the call.
Information includes interchange features, linear reference markers, and directionalities.
“Incident”“10–50/51”, “PI (Personal Injury)”, “crash”, “collision”, “wreck”, “fire”, “medical”Common coded communications and contextual information about the incident contained within the call.
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Gartner, C.M.; Vajpayee, V.; Desai, J.; Bullock, D.M. Automatic Speech Recognition of Public Safety Radio Communications for Interstate Incident Detection and Notification. Smart Cities 2025, 8, 157. https://doi.org/10.3390/smartcities8050157

AMA Style

Gartner CM, Vajpayee V, Desai J, Bullock DM. Automatic Speech Recognition of Public Safety Radio Communications for Interstate Incident Detection and Notification. Smart Cities. 2025; 8(5):157. https://doi.org/10.3390/smartcities8050157

Chicago/Turabian Style

Gartner, Christopher M., Vihaan Vajpayee, Jairaj Desai, and Darcy M. Bullock. 2025. "Automatic Speech Recognition of Public Safety Radio Communications for Interstate Incident Detection and Notification" Smart Cities 8, no. 5: 157. https://doi.org/10.3390/smartcities8050157

APA Style

Gartner, C. M., Vajpayee, V., Desai, J., & Bullock, D. M. (2025). Automatic Speech Recognition of Public Safety Radio Communications for Interstate Incident Detection and Notification. Smart Cities, 8(5), 157. https://doi.org/10.3390/smartcities8050157

Article Metrics

Back to TopTop