Recognition Performance of Selected Speech Recognition APIs – A Longitudinal Study

Siegert, Ingo; Sinha, Yamini; Jokisch, Oliver; Wendemuth, Andreas

doi:10.1007/978-3-030-60276-5_50

Cited by 5 publications

(4 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To find the best cloud-based speech recognition service, Ingo Siegert et al [9] analyzed the performance of Google Speech API [19], IBM Watson [20], and Wit.ai [21] on German samples of high-quality spontaneous human-directed and devicedirected speech, as well as noisy device-directed speech, over an eight-month period. The experimental results [18] have shown that word error rate is quite low in Google Cloud Speech and Wit.ai with 7% to 8% absolute.…”

Section: Openvidu-call Jitsi Meet Bigbluebuttonmentioning

confidence: 95%

“…Amazon Alexa API, Nuance Recognizer, Wit.ai, Houndify), some provide the possibility to use their speech recognition API also within personal products or with own speech files [18]. To find the best cloud-based speech recognition service, Ingo Siegert et al [9] analyzed the performance of Google Speech API [19], IBM Watson [20], and Wit.ai [21] on German samples of high-quality spontaneous human-directed and devicedirected speech, as well as noisy device-directed speech, over an eight-month period.…”

Section: Openvidu-call Jitsi Meet Bigbluebuttonmentioning

confidence: 99%

See 1 more Smart Citation

Real-Time Facial Expression Recognition and Speech Tran-scripts over an on-premise Video Conference Application

Ahmed

Areed

Obayya

et al. 2022

International Journal of Telecommunications

View full text Add to dashboard Cite

Since Covid-19 pandemic outbreak, organizations and individuals have had to use video conference applications increasingly. However, the commercial video conference applications are expensive, and feature limited. This paper discusses how to enable organizations to host onpremise video conference applications. Then, it explores assisting organization's stakeholders with making decisions based on facial expressions of video conference attendees. Moreover, it facilitates transcribing speech into text to enable deaf persons to participate in online conferences. Technologies and tools used in addressing these challenges respectively are: (i) Web Real Time Communication (WebRTC) project, (ii) Tensorflow.js library, (iii) and Web Speech Application Programming Interface (API). This paper depends on integration between a collection of technologies, libraries, standards, and protocols. Most of them can be managed using JavaScript framework. Hence, load of the performance is distributed on each client-side device. The proposed onpremise video conference application has been enhanced through including facial expression recognition with 66% high accuracy while the speech-into-text feature with Word Error Rates (WER) are 0 and 0.12 for British English and Egyptian Arabic, respectively.

show abstract

Section: Openvidu-call Jitsi Meet Bigbluebuttonmentioning

confidence: 95%

Section: Openvidu-call Jitsi Meet Bigbluebuttonmentioning

confidence: 99%

Real-Time Facial Expression Recognition and Speech Tran-scripts over an on-premise Video Conference Application

Ahmed

Areed

Obayya

et al. 2022

International Journal of Telecommunications

View full text Add to dashboard Cite

show abstract

“…Though in the following years, Hebrew ASR performance may have improved, we expect that the WER of Hebrew will still be higher than typical WERs of English. For German, Siegert et al (2020) reported 7% WER. Hence, we hypothesize: H1: A comparison of English, Hebrew, and German ASR would yield the highest performance for English, the lowest performance for Hebrew, and an in-between performance for German.…”

Section: Rq1mentioning

confidence: 99%

“…The purpose of this research is to explore ASR performance gaps by a comparative study of American English (henceforth, English), which is a high-resourced language, versus German and Hebrew as a mid-resourced and low-resourced language, respectively. Performance evaluation is a major issue in scientific research and technology development (NIST, 2009;Siegert et al, 2020).…”

Section: Introductionmentioning

confidence: 99%

A cross-language study of speech recognition systems for English, German, and Hebrew

Silber-Varod

Siegert

Jokisch

et al. 2021

OJAKM

Self Cite

View full text Add to dashboard Cite

Despite the growing importance of Automatic Speech Recognition (ASR), its application is still challenging, limited, language-dependent, and requires considerable resources. The resources required for ASR are not only technical, they also need to reflect technological trends and cultural diversity. The purpose of this research is to explore ASR performance gaps by a comparative study of American English, German, and Hebrew. Apart from different languages, we also investigate different speaking styles – utterances from spontaneous dialogues and utterances from frontal lectures (TED-like genre). The analysis includes a comparison of the performance of four ASR engines (Google Cloud, Google Search, IBM Watson, and WIT.ai) using four commonly used metrics: Word Error Rate (WER); Character Error Rate (CER); Word Information Lost (WIL); and Match Error Rate (MER). As expected, findings suggest that English ASR systems provide the best results. Contrary to our hypothesis regarding ASR’s low performance for under-resourced languages, we found that the Hebrew and German ASR systems have similar performance. Overall, our findings suggest that ASR performance is language-dependent and system-dependent. Furthermore, ASR may be genre-sensitive, as our results showed for German. This research contributes a valuable insight for improving ubiquitous global consumption and management of knowledge and calls for corporate social responsibility of commercial companies, to develop ASR under Fair, Reasonable, and Non-Discriminatory (FRAND) terms

show abstract

Simulation Architecture

Michael

2023

Simulating Conversations for the Prediction of Speech Quality

View full text Add to dashboard Cite

Recognition Performance of Selected Speech Recognition APIs – A Longitudinal Study

Cited by 5 publications

References 11 publications

Real-Time Facial Expression Recognition and Speech Tran-scripts over an on-premise Video Conference Application

Real-Time Facial Expression Recognition and Speech Tran-scripts over an on-premise Video Conference Application

A cross-language study of speech recognition systems for English, German, and Hebrew

Simulation Architecture

Contact Info

Product

Resources

About