Datasets:

sakshee05
/

alphanumeric-audio-dataset

Response_ID stringlengths 17 17	Age float64 20 39 ⌀	Gender stringclasses 3 values	Nationality stringclasses 10 values	Native Language stringclasses 16 values	Familiarity with English stringclasses 3 values	Accent Strength (Self reported) int64 0 9	Difficulties stringclasses 4 values	Recording Machine stringclasses 3 values	Name stringlengths 5 25	Number int64 7 9.9B	Address stringlengths 15 50	Duration (secs) int64 110 984
R_14L9qNxUlipFmsF	23	Male	Indian	Tamil	Educated in English, but not my native language	7	Sometimes	Laptop	Alina Patel	3,041,254,855	107 Crystal Lake Drive 327	399
R_15Rh8IA5GvyzCIF	23	Female	Indian	Marathi	Educated in English, but not my native language	5	Sometimes	Phone Recorder	Jarek Ferrari	1,045,562,375	84 River Road 814	111
R_1DN3Yy2ILEefmz8	24	Female	Indian	Hindi	Educated in English, but not my native language	4	Sometimes	Phone Recorder	Seraphina Balasubramanian	43,126,798	99 Willow Lane 493	292
R_1KvCEEGNThnpnzy	24	Female	Indian	Marathi	Educated in English, but not my native language	5	About half the time	Phone Recorder	Saffron Robles	8,885,112,931	48 Ridge Road 407	278
R_1O7z12FZh2bL6ST	22	Male	American	English	Native speaker	0	Sometimes	Phone Recorder	Zane Robles	5,993,415,111	34 Elm Street 568	509
R_1QrO96YnPAaHofK	26	Prefer not to say	Indian	Kannada	Basic knowledge; not fluent	5	About half the time	Phone Recorder	Ivan Hwang	1,928,374,655	430 Autumn Lane 274	136
R_1VaNGgPhYxnUTbC	25	Male	Indian	Gujarati	Educated in English, but not my native language	7	Sometimes	Phone Recorder	Niklas Shah	5,142,296,612	86 Timberline Boulevard 664	232
R_1b0CnNoTm0uIas5	23	Male	Indian	Malayalam	Educated in English, but not my native language	2	Sometimes	Phone Recorder	Ravi Choi	8,292,159,357	48 Ridge Road 407	353
R_1bhWwiBjHlA1GcF	null	Male	Indo-canadian	Hindi	Educated in English, but not my native language	6	Sometimes	Phone Recorder	Eric dundas	7	1315 morreene road	536
R_1oLUlgQk6qINvsk	27	Female	Indian	Kannada	Basic knowledge; not fluent	7	About half the time	Phone Recorder	Roshan Coelho	6,163,483,977	77 Maplewood Drive 659	152
R_1poePVg1gzUKUz7	24	Male	North African	Arabic	Educated in English, but not my native language	4	Sometimes	Laptop	Juniper Leung	9,679,778,449	60 Pebble Beach Road 743	333
R_1tDKM2lNpA5vaZb	25	Female	Indian	Kannada	Educated in English, but not my native language	1	Never	Phone Recorder	Sabrina Shirodkar	9,568,214,730	114 Park Avenue, Chicago, IL, 67004	741
R_1wB3pJ1YQM5wyqI	21	Male	American	English	Native speaker	0	Sometimes	Phone Recorder	Seraphina Ivanov	2,288,775,400	128 Cedar Avenue 401	252
R_1zZKZdkb4iInKK7	23	Male	Chinese	Chinese	Educated in English, but not my native language	7	Sometimes	Laptop	Malia Tanaka	9,281,822,959	21 Harbor View Drive 706	372
R_1zin13bSMyOVRnJ	25	Female	Chinese	Chinese	Educated in English, but not my native language	3	Sometimes	Phone Recorder	Layla Knudsen	5,363,648,030	8 Horizon Avenue 613	503
R_34Kh357EGklz2X7	27	Male	Indian	Hindi	Educated in English, but not my native language	4	Sometimes	Laptop	Sable Daniel	2,488,140,980	58 Maple Valley Road 812	372
R_34RasWwKijJI5sB	22	Female	Indian	Hindi	Educated in English, but not my native language	5	Most of the time	Phone Recorder	Calder Fuchs	1,855,749,866	160 Sunset Drive 885	114
R_34i05S8YXzyAyGE	26	Female	Argentinian	Spanish	Educated in English, but not my native language	5	Sometimes	Phone Recorder	Isla Santos	6,537,986,847	55 Stonewall Street 335	852
R_3OihBnBKQuMFAz4	21	Female	Indian	Tamil	Educated in English, but not my native language	4	Most of the time	Phone Recorder	Aurora Ferrari	9,299,574,124	77 Spruce Way 920	174
R_3f5pnhlvLlxkfu1	28	Male	Argentinian	Spanish	Basic knowledge; not fluent	7	Most of the time	Phone Recorder	Elysia Lobo	2,615,640,089	14 Silver Street 964	529
R_3g4JxW8aBdFTYoc	23	Female	Indian	Tamil	Educated in English, but not my native language	7	About half the time	Phone Recorder	Laila Zhou	5,116,853,235	32 Crystal Springs Boulevard 450	134
R_3gToFAf4MoM9dl8	22	Female	Indian	Marathi	Educated in English, but not my native language	4	Sometimes	Phone Recorder	Rami Vasilyev	7,499,218,381	99 Sunnyville Boulevard 342	180
R_3i2ltOSffNPHRfR	22	Male	American	English	Native speaker	0	Never	Phone Recorder	Ciaran MacLeod	2,853,006,411	5 Ocean Drive 432	339
R_3nAeHdtYMOFbRQJ	21	Female	Indian	Hindi	Educated in English, but not my native language	3	Sometimes	Phone Recorder	Anannya	9,196,856,019	7204, McQueen Drive	542
R_3uvOS0cUzHJeF6p	24	Male	Indian	Hindi	Basic knowledge; not fluent	5	Sometimes	Phone Recorder	Niko Coelho	3,053,578,003	75 Valley Grove Road 768	146
R_3vhWAQpPNAgLass	22	Male	Indian	Telugu	Educated in English, but not my native language	8	Sometimes	Laptop	Rowena Taneja	4,127,415,553	456 Birch Boulevard 349	266
R_41aAifnTBI6jslp	20	Female	Indian	Hindi	Educated in English, but not my native language	9	Sometimes	Phone Recorder	Jorah Morment	5,286,564,231	92 Dragos Khalasar Dothrakhi	401
R_41aiHA1p9OLYbuY	26	Male	Indian	Marathi	Educated in English, but not my native language	2	Sometimes	Phone Recorder	Ranjendra Pingle	1,234,567,898	Flat 202 Roop Ganga Pune	489
R_49dhghc8TaDIOhr	20	Female	Indian	Hindi	Educated in English, but not my native language	8	Sometimes	Phone Recorder	Arya Stark	1,253,579,821	37 Casterly Rock Lannis Port	480
R_4Cbjs7mWxj4Ri7v	20	Female	Indian	Hindi	Educated in English, but not my native language	7	Most of the time	Phone Recorder	Aditi Rampiya	7,998,735,821	Lane Number 4, Swimming Pool	258
R_4GQeQB9GOpNtRaW	20	Female	Indian	Hindi	Educated in English, but not my native language	7	Sometimes	Phone Recorder	Taklu Analogs	7,413,825,402	741 Opamps Circuit Lane, Barkha Avenue, Tiwari	255
R_4JsyJD7ticKbRiZ	29	Female	Indian	Hindi	Basic knowledge; not fluent	8	Sometimes	Phone Recorder	Darshana	123,456,789	FLAT NO 204 SATSANG SOCIETY THANE MUMBAI 411007	499
R_4M4K1OEOS6Dy9xr	35	Male	Indian	Marathi	Educated in English, but not my native language	7	Sometimes	Phone Recorder	Ganesh Kathale	813,546,798	FLAT NO 701 RAGADARI APARTMENT NASHIK 423440	271
R_4QJpbtn4T8z4ixS	25	Male	Indian	Marathi	Educated in English, but not my native language	8	Sometimes	Laptop	Jitu Shinde	7,773,330,000	FLAT NO 1100 SATSANG COLONY SHAHADA 450089	149
R_4RQSv9iOZkB2f9Z	24	Female	Indian	Marathi	Educated in English, but not my native language	5	Sometimes	Phone Recorder	Hiroshi Khatri	4,001,291,474	10 Woodland Drive 748	397
R_4TIdkLil27HhMH3	22	Female	Indian	Hindi	Basic knowledge; not fluent	8	Sometimes	Phone Recorder	Shubhangi katkar	9,530,589,346	Flat no 303 E L Homes Baner pune 411007	304
R_4aYapHeUHBuU8XG	24	Female	Indian	Marathi	Educated in English, but not my native language	6	Sometimes	Phone Recorder	Soraya Balasubramanian	2,312,324,633	110 Bluebell Lane 973	240
R_4ezaMzei6ssNagt	23	Male	Indian	Marathi	Educated in English, but not my native language	8	Sometimes	Phone Recorder	Vikas Raj	6,578,902,300	FLAT NO 4 RAMBAG SOCIETY BARUCH 390345	201
R_4gezxpDDRaPXKIs	30	Male	Indian	Gujrati	Educated in English, but not my native language	9	Sometimes	Laptop	Prashant Gandhe	123,459,876	PLOT NO 30 PRATIK COLONY SURAT 356003	663
R_4kJncQejh6M9Qxb	20	Female	Indian	Hindi	Educated in English, but not my native language	8	Sometimes	Phone Recorder	Battli Aroda	2,512,254,291	69 Cubicle, Bathroom Street, UK	194
R_4s7PwXExwuw1cZa	32	Male	Indian	Hindi	Educated in English, but not my native language	8	Sometimes	Phone Recorder	Arjun	987,654,321	Plot No 43 Pasaydan Profeesor colony Jagaon 424005	251
R_4shv1y28wS6SaPv	29	Female	Indian	Hindi	Educated in English, but not my native language	4	Never	Phone Recorder	Mehak	9,811,219,666	Fenway park, boston, Massachusetts	577
R_4ssmicMjtQagK7v	39	Male	Indian	Marathi	Educated in English, but not my native language	7	Sometimes	Phone Recorder	Mayur Shah	9,898,005,600	FLAT NO 207 SNEHA VIHAR KOLHAPUR 411002	787
R_4zatSIgrUgsETE5	23	Female	Indian	Marathi	Basic knowledge; not fluent	8	Sometimes	Phone Recorder	Suman patidar	9,370,524,963	Swami sandipani coloney solapur	310
R_51hzd8Q0PcCS4Jb	22	Male	Indian	Hindi	Educated in English, but not my native language	4	Sometimes	Phone Recorder	Harlow Qureshi	8,829,090,933	290 Seaside Boulevard 275	394
R_57WCdr3dpzLNhbe	26	Female	Chinese Canadian	Mandarin	Educated in English, but not my native language	5	Never	Phone Recorder	Liam Ali	8,114,844,717	3 Meadowbrook Road 874	211
R_5F0Zpg07Mr16uBz	25	Female	Indian	English	Native speaker	5	Never	Phone Recorder	Harlow Smith	5,595,476,016	45 Ivy Lane 602	436
R_5MRLVafIukImdof	23	Male	American	Chinese	Educated in English, but not my native language	1	Sometimes	Phone Recorder	Saffron Abraham	3,788,618,965	5 Ocean Drive 432	254
R_5MYVgvg6SSesm0n	28	Female	Indian	Hindi	Educated in English, but not my native language	6	Sometimes	Phone Recorder	Maxine Jensen	1,056,058,263	40 Westfield Road 207	122
R_5hBj9gxvIoDecnz	23	Female	American	Korean	Educated in English, but not my native language	3	Sometimes	Phone Recorder	Elise Taylor	7,285,242,379	49 Windy Ridge Avenue 240	310
R_5jebEVEgcwRTOcO	23	Prefer not to say	Indian	Telugu	Educated in English, but not my native language	3	Sometimes	Phone Recorder	Calder Jackson	1,808,231,222	301 Massachusetts Avenue 1462	140
R_5lWkQMcCDfrgWwV	25	Male	Indian	Hindi	Educated in English, but not my native language	8	Sometimes	Phone Recorder	Liam Patel	9,783,812,441	52 Golden Ridge Drive 418	244
R_5re0w7ElTMqvkCP	23	Female	Indian	Kannada	Educated in English, but not my native language	3	Never	Phone Recorder	Beatriz Kim	4,904,368,573	207 Westside Lane 528	151
R_6CxrKuxhJ3GtN8K	23	Male	Chinese Canadian	Mandarin	Educated in English, but not my native language	1	Sometimes	Phone Recorder	Jon Reifshneider	1,234,567,890	8 Horizon Avenue 613	388
R_6E6ELKr8L1H0RnH	23	Female	Chinese	Chinese	Educated in English, but not my native language	2	Sometimes	Phone Recorder	Calla Pereira	1,709,916,637	3 Meadowbrook Road 874	972
R_6pnzU40E9DSpr61	23	Male	American	English	Native speaker	0	Sometimes	Phone Recorder	Lennox Anderson	3,619,713,354	11438 SE 184th Pl, Renton WA	601
R_6rCTqJtaK9fNxHq	22	Female	Indian	Hindi	Educated in English, but not my native language	6	Sometimes	Phone Recorder	Isla de Souza	9,661,113,443	52 Golden Ridge Drive 418	142
R_6thxkopa8CNYELv	24	Male	American	English	Native speaker	7	Never	Phone Recorder	Ziya Ali	8,178,069,680	202 Northgate Drive 512	233
R_77p04jkohla1sH9	24	Female	Chinese	Chinese	Educated in English, but not my native language	7	Never	Phone Recorder	Celine Gupta	1,712,391,906	75 Pine Mountain Road 705	984
R_7DHnUtjhzr5gTvl	24	Male	Indian	Hindi	Educated in English, but not my native language	5	Never	Phone Recorder	Elysia Peinado	5,473,978,936	150 Riverbend Avenue 465	359
R_7EOlSvi9ctgFRBL	23	Male	Chinese	Chinese	Educated in English, but not my native language	2	Sometimes	Phone Recorder	Norah Cho	9,356,122,945	75 Pine Mountain Road 705	928
R_7H8pYLcVGrTM7g5	23	Female	Indo-English	English	Native speaker	1	Never	Phone Recorder	Freya Hamed	1,314,559,375	78 Sunset Boulevard 789	112
R_7HEfVLTSyuEezdf	28	Female	Indian	Hindi	Educated in English, but not my native language	4	Sometimes	Phone Recorder	Sanaa Letourneau	6,686,412,385	67 Mountain View Drive 214	120
R_7NCxNP8jSj5mq6n	26	Female	Indian	Hindi	Educated in English, but not my native language	6	Sometimes	Phone Recorder	Zander Burrows	4,885,080,040	50 Harvest Lane 591	195
R_7OVSal7s7tNbjLO	24	Male	Indian	Hindi	Educated in English, but not my native language	5	About half the time	Laptop	Asher Singh	8,717,379,417	161 Fairview Street 407	276
R_7QGd13jsljnaCbY	25	Female	Indian	Telugu	Educated in English, but not my native language	5	Sometimes	Phone Recorder	Harshitha Rasamsetty	2,873,764,800	44 Sand Dune Road 205	221
R_7Vlj2RdVyGOM7Yz	25	Female	Indian	Urdu	Educated in English, but not my native language	5	Most of the time	Phone Recorder	Solana Chatterjee	5,735,062,913	29 Summer Breeze Lane 571	110
R_7iI7wmwsdCo4PbX	22	Female	American	Chinese	Educated in English, but not my native language	7	Sometimes	Phone Recorder	Hina Karpov	8,985,955,886	15 Sea Breeze Avenue 823	243
R_7pKNTfB39Yc3I41	23	Male	Middle Asian	Uyghur	Educated in English, but not my native language	2	Sometimes	Phone Recorder	Benji Chatterjee	6,341,735,686	128 Cedar Avenue 401	509
R_7qeKPomI8VjGOsK	29	Male	American	English	Native speaker	2	Sometimes	External Microphone	Zuri Sutherland	2,563,144,191	56 Pine Tree Lane 640	590
R_14L9qNxUlipFmsF	23	Male	Indian	Tamil	Educated in English, but not my native language	7	Sometimes	Laptop	Alina Patel	3,041,254,855	107 Crystal Lake Drive 327	399
R_15Rh8IA5GvyzCIF	23	Female	Indian	Marathi	Educated in English, but not my native language	5	Sometimes	Phone Recorder	Jarek Ferrari	1,045,562,375	84 River Road 814	111
R_1DN3Yy2ILEefmz8	24	Female	Indian	Hindi	Educated in English, but not my native language	4	Sometimes	Phone Recorder	Seraphina Balasubramanian	43,126,798	99 Willow Lane 493	292
R_1KvCEEGNThnpnzy	24	Female	Indian	Marathi	Educated in English, but not my native language	5	About half the time	Phone Recorder	Saffron Robles	8,885,112,931	48 Ridge Road 407	278
R_1O7z12FZh2bL6ST	22	Male	American	English	Native speaker	0	Sometimes	Phone Recorder	Zane Robles	5,993,415,111	34 Elm Street 568	509
R_1QrO96YnPAaHofK	26	Prefer not to say	Indian	Kannada	Basic knowledge; not fluent	5	About half the time	Phone Recorder	Ivan Hwang	1,928,374,655	430 Autumn Lane 274	136
R_1VaNGgPhYxnUTbC	25	Male	Indian	Gujarati	Educated in English, but not my native language	7	Sometimes	Phone Recorder	Niklas Shah	5,142,296,612	86 Timberline Boulevard 664	232
R_1b0CnNoTm0uIas5	23	Male	Indian	Malayalam	Educated in English, but not my native language	2	Sometimes	Phone Recorder	Ravi Choi	8,292,159,357	48 Ridge Road 407	353
R_1bhWwiBjHlA1GcF	null	Male	Indo-canadian	Hindi	Educated in English, but not my native language	6	Sometimes	Phone Recorder	Eric dundas	7	1315 morreene road	536
R_1oLUlgQk6qINvsk	27	Female	Indian	Kannada	Basic knowledge; not fluent	7	About half the time	Phone Recorder	Roshan Coelho	6,163,483,977	77 Maplewood Drive 659	152
R_1poePVg1gzUKUz7	24	Male	North African	Arabic	Educated in English, but not my native language	4	Sometimes	Laptop	Juniper Leung	9,679,778,449	60 Pebble Beach Road 743	333
R_1tDKM2lNpA5vaZb	25	Female	Indian	Kannada	Educated in English, but not my native language	1	Never	Phone Recorder	Sabrina Shirodkar	9,568,214,730	114 Park Avenue, Chicago, IL, 67004	741
R_1wB3pJ1YQM5wyqI	21	Male	American	English	Native speaker	0	Sometimes	Phone Recorder	Seraphina Ivanov	2,288,775,400	128 Cedar Avenue 401	252
R_1zZKZdkb4iInKK7	23	Male	Chinese	Chinese	Educated in English, but not my native language	7	Sometimes	Laptop	Malia Tanaka	9,281,822,959	21 Harbor View Drive 706	372
R_1zin13bSMyOVRnJ	25	Female	Chinese	Chinese	Educated in English, but not my native language	3	Sometimes	Phone Recorder	Layla Knudsen	5,363,648,030	8 Horizon Avenue 613	503
R_34Kh357EGklz2X7	27	Male	Indian	Hindi	Educated in English, but not my native language	4	Sometimes	Laptop	Sable Daniel	2,488,140,980	58 Maple Valley Road 812	372
R_34RasWwKijJI5sB	22	Female	Indian	Hindi	Educated in English, but not my native language	5	Most of the time	Phone Recorder	Calder Fuchs	1,855,749,866	160 Sunset Drive 885	114
R_34i05S8YXzyAyGE	26	Female	Argentinian	Spanish	Educated in English, but not my native language	5	Sometimes	Phone Recorder	Isla Santos	6,537,986,847	55 Stonewall Street 335	852
R_3OihBnBKQuMFAz4	21	Female	Indian	Tamil	Educated in English, but not my native language	4	Most of the time	Phone Recorder	Aurora Ferrari	9,299,574,124	77 Spruce Way 920	174
R_3f5pnhlvLlxkfu1	28	Male	Argentinian	Spanish	Basic knowledge; not fluent	7	Most of the time	Phone Recorder	Elysia Lobo	2,615,640,089	14 Silver Street 964	529
R_3g4JxW8aBdFTYoc	23	Female	Indian	Tamil	Educated in English, but not my native language	7	About half the time	Phone Recorder	Laila Zhou	5,116,853,235	32 Crystal Springs Boulevard 450	134
R_3gToFAf4MoM9dl8	22	Female	Indian	Marathi	Educated in English, but not my native language	4	Sometimes	Phone Recorder	Rami Vasilyev	7,499,218,381	99 Sunnyville Boulevard 342	180
R_3i2ltOSffNPHRfR	22	Male	American	English	Native speaker	0	Never	Phone Recorder	Ciaran MacLeod	2,853,006,411	5 Ocean Drive 432	339
R_3nAeHdtYMOFbRQJ	21	Female	Indian	Hindi	Educated in English, but not my native language	3	Sometimes	Phone Recorder	Anannya	9,196,856,019	7204, McQueen Drive	542
R_3uvOS0cUzHJeF6p	24	Male	Indian	Hindi	Basic knowledge; not fluent	5	Sometimes	Phone Recorder	Niko Coelho	3,053,578,003	75 Valley Grove Road 768	146
R_3vhWAQpPNAgLass	22	Male	Indian	Telugu	Educated in English, but not my native language	8	Sometimes	Laptop	Rowena Taneja	4,127,415,553	456 Birch Boulevard 349	266
R_41aAifnTBI6jslp	20	Female	Indian	Hindi	Educated in English, but not my native language	9	Sometimes	Phone Recorder	Jorah Morment	5,286,564,231	92 Dragos Khalasar Dothrakhi	401
R_41aiHA1p9OLYbuY	26	Male	Indian	Marathi	Educated in English, but not my native language	2	Sometimes	Phone Recorder	Ranjendra Pingle	1,234,567,898	Flat 202 Roop Ganga Pune	489
R_49dhghc8TaDIOhr	20	Female	Indian	Hindi	Educated in English, but not my native language	8	Sometimes	Phone Recorder	Arya Stark	1,253,579,821	37 Casterly Rock Lannis Port	480
R_4Cbjs7mWxj4Ri7v	20	Female	Indian	Hindi	Educated in English, but not my native language	7	Most of the time	Phone Recorder	Aditi Rampiya	7,998,735,821	Lane Number 4, Swimming Pool	258

End of preview. Expand in Dataset Viewer.

Speech Recognition Bias Reduction Project

Executive Summary

Welcome to the Speech Recognition Bias Reduction Project. It aims to create a more inclusive and representative dataset for improving automated speech recognition systems. This project addresses the challenges faced by speakers with non-native English accents, particularly when interacting with automated voice systems that struggle to interpret alphanumeric information such as names, phone numbers, and addresses.

Motivated by my personal experience as an international student moving to the United States in 2024, I aim to reduce delays, misinterpretations, and ineffective communication caused by accent-based biases in current voice recognition technologies.

Objective

The primary goal is to create a diverse dataset of alphanumeric audio inputs. This dataset will focus on audio recordings from speakers with various accents, starting with the vibrant community of international students at Duke University.

It aims to tackle -
Dataset Bias: Reduce native English bias and make voice recognition systems more inclusive.
Efficiency/Equality: In a multicultural society, ensuring equality in access to services is crucial.

Potential Applications

Improving Voice Recognition: Make speech systems better at understanding accents, especially when it comes to things like spelling out names or reading phone numbers.
Reducing Bias in AI: Help make voice recognition tech more inclusive by reducing bias against non-native English speakers.
Linguistic Insights: Provide data to understand how different accents impact the way people say things like phone numbers and addresses.
Language Learning & Accessibility: Support tools for language learners or accessibility projects, helping AI understand a wider range of speech patterns.

Review of Previous Datasets

Existing audio datasets primarily focus on word and sentence data to enhance representation of diverse accents: Common Voice, VoxCeleb, LibriSpeech. Datasets focusing on “Phonetics” are

https://catalog.ldc.upenn.edu/LDC93S1: Focuses on American English dialects, which may not adequately represent non-native speakers.
https://archive.phonetics.ucla.edu/: An extensive archive that could be utilized to up sample the collected data.
https://en.arabicspeechcorpus.com/ : Valuable for Arabic language processing, but it doesn’t address English accents.
https://github.com/S-Malek/PCVC: focuses on Modern Persian speech, lacking relevance for English accents.

Novelty

Focus on Alphanumeric Data: The dataset specifically targets the recognition of letters and numbers, critical for automated systems that handle personal information.
Diverse Accent Representation: The dataset will prioritize non-native English speakers, enabling a nuanced understanding of how various accents influence recognition accuracy.
Rich Metadata: Detailed demographic information allows for in-depth analysis and helps identify patterns in recognition challenges faced by different accent groups.

Dataset Description and Collection Protocol / Tools Used

Please refer docs/collection_protocol.md and docs/dataset_description.md.

Power Analysis

The calculated sample size is approximately 36 participants per group.

For Group 1, there are 55 data points fulfilling the power analysis expectation. However for Group 2, the data falls short of 29 participants.

Exploratory Data Analysis

The age distribution of participants skews towards the 20–30 age range, reflecting the demographic characteristics of the survey population, which predominantly consists of university students. However, this limited range does introduce age representation bias.

The gender distribution in the dataset is balanced ensuring that models trained on the dataset can generalize reasonably well across genders.

Both nationality and native language distributions reveal noticeable representation bias. A substantial proportion of participants hail from India, likely due to the my community access. This concentration also results in overrepresentation of Hindi (India's national language) and Marathi (my native language). This bias limits the diversity of accents in the dataset. Future users of the dataset may need to augment or balance the data to achieve more equitable representation of accents and linguistic backgrounds.

Despite the class imbalance in the "Familiarity with English" distribution, it effectively captures the necessary data for the intended study, with each group represented. The imbalance is likely a cascading effect of the nationality distribution, as the majority of participants are Indian students who are typically educated in English alongside their native language.

Survey Completion Insights

The average time to complete the survey was approximately 5.9 minutes, indicating that the process was relatively streamlined. However, platform limitations in Qualtrics, which lacks native audio recording capabilities, required participants to record and upload files separately. This additional effort likely contributed to lower participation rates, demonstrating the importance of minimizing participant burden in data collection to improve response rates.

Ethics Statement

The Alphanumeric Audio Dataset was collected with strict adherence to ethical guidelines:

Informed Consent: Participants were fully informed about the purpose and use of their contributions, with consent obtained before participation.
Anonymization: All data was anonymized to protect participant privacy, with only non-identifiable metadata included for analysis.
IRB Approval: The study was reviewed and approved by Duke University's Institutional Review Board (IRB) to ensure compliance with ethical research standards.
Voluntary Participation: Participation was voluntary, and participants could withdraw their data before public release.
Responsible Use: The dataset is open-sourced to promote inclusivity in AI research and must be used for ethical, non-discriminatory purposes.

License

This dataset is licensed under the MIT License.

You are free to use, modify, and distribute this dataset for any purpose, including commercially, as long as you include the original copyright notice in all copies or substantial portions of the dataset. The dataset is provided "as is", without warranty of any kind.

Contact Information

For any questions or further information, please contact:

Sakshee Patil
Email: sakshee.patil@duke.edu

** Few Sections of the README are re-articulated using ChatGPT.

Downloads last month: 398

Size of downloaded dataset files:

192 MB

Size of the auto-converted Parquet files:

181 MB

Number of rows:

210