Science topic

ASR - Science topic

Explore the latest questions and answers in ASR, and find ASR experts.
Questions related to ASR
  • asked a question related to ASR
Question
4 answers
Hello, I've recently been studying Ancestral Sequence Reconstruction (ASR), attempting to infer ancestral sequences of viruses. I understand that this inference is constrained by factors like sample size and models, and represents a plausible sequence that may have existed. However, I'm curious about whether directly comparing these inferred ancestral sequences holds biological significance. Can they reflect the differences among the extant sequences from various lineages that were used to infer them?
Relevant answer
Answer
Hongzhuang Chen I am afraid that you can lose a lot of information from such comparison. But, it can be applied (and very useful) to illustrate the differences supported statistically by analysis of the original data (sequences).
  • asked a question related to ASR
Question
2 answers
I am interested in publishing research on deep learning-based automatic speech recognition (ASR) for a specific low-resource language. What are the key areas or research gaps that I should focus on to contribute to this field? Are there any influential papers or research works that I should read to gain a comprehensive understanding of the current state-of-the-art in this area? I would greatly appreciate any guidance or recommendations from experts in this field. Thank you in advance!
Relevant answer
Answer
The research gaps include data scarcity, lack of text transcriptions, challenges with out-of-vocabulary words, and handling accent and dialect variability. Additionally, code-switching and multilingualism, model adaptation, and robustness to adverse acoustic conditions are important areas for exploration.
You can focus on developing novel data augmentation techniques, effective transfer learning and multilingual modeling, and exploring unsupervised and semi-supervised learning methods. Investigate robustness to adverse acoustic conditions, devise language-specific modeling approaches, and work on active learning and data collection strategies to optimize the annotation process.
  • asked a question related to ASR
Question
2 answers
I need to calculate oxygen exchange current density using polarization resistance.
Relevant answer
Answer
Dear Sunita Saharan thank you so much for your response. Here I want to convey one thing in many papers ASR= Rp* Surface area of electrode/2. For reference, I attached a file below. So my doubt is if the unit of Rp is ohm.cm^2 then is it possible that unit of ASR will also be ohm.cm^2. Please clarify this.
Thank you
  • asked a question related to ASR
Question
1 answer
area specific resistance (ASR) in SOFC
Relevant answer
Answer
Area specific resistance is the resistance offered by unit area of the cell. Can be calculated by dividing the total thickness of the MEA by effective conductivity of the MEA.
The unit is Ohms-m2
  • asked a question related to ASR
Question
10 answers
I need to analyze some qualitative data including both video and voice recording for my master thesis. Before the process of data analysis, I will need the transcripts (written forms) of these qualitative data. Therefore, I am in need of some software or websites having the function of automated speech recognition (ASR) system. I will be appreciated if it is free of charge.
Looking forward to hearing some suggestions from senior researchers.
Relevant answer
Answer
Thank you for the information you have provided. Nishat Vasker I will check them out.
  • asked a question related to ASR
Question
4 answers
I am trying to find ASR/ACR potential of micritic limestone when used as a coarse aggregate in asphalt and concrete.
Relevant answer
Answer
ASR is Alkalia - Silica Reactivity. It is a reaction between the surface of silica based rocks, and Portland Cement. Substituting about 20 % Fly Ash from coal fired power plants ( for the Portland Cement ) severely reduces this gel forming expansive reaction. Since Limestone does not have silica in it, and limestone is one of the main components of Portland Cement, there should be no Alkalia - Silica reaction. Limestone is Calcium Carbonate (Ca CO3), there is no SiO2.
  • asked a question related to ASR
Question
4 answers
The alkali-silica reaction (ASR), known as "concrete cancer" is a deleterious swelling reaction that occurs over time in concrete between the highly alkaline cement paste and the reactive amorphous (i.e., non-crystalline) silica found in many common aggregates, given sufficient moisture. What are the emerging repair and prevention techniques for ASR?.
Relevant answer
Answer
Utilization of pozzolanic materials in concrete prevents alkali silica reaction by controlling and limiting lime (calcium hydroxide) content through pozzolanic reaction to produce more calcium silicate/aluminate hydrates (CSH or CAH).
  • asked a question related to ASR
Question
3 answers
I am working on ASR creation for an under-resourced Indo-Aryan languange. As a start I want to work on isolated word recognition. I will be requiring phonetic dictionary as part of this work. can you suggest me some good reading materials for automatic phonetic dictionary creation.
Relevant answer
Answer
You could check out POnSS as reported in the following study. Rodd, J., Decuyper, C., Bosker, H. R., & Ten Bosch, L. (2020). A tool for efficient and accurate segmentation of speech data: Announcing POnSS. Behavior Research Methods, 53(2), 744-756. https://doi.org/10.3758/s13428-020-01449-6 Good luck,
  • asked a question related to ASR
Question
5 answers
Can anyone help me with finding a dataset for alkali-silica reaction(ASR) cracks? Thanks
Relevant answer
Answer
  • asked a question related to ASR
Question
2 answers
Features extracted using MFCC gives 13*n matrix as the extracted features. Whereas those extracted using GFCC gives a 64*n matrix.
What is the reason for this?
https://github.com/mvansegbroeck/featxtra/issues/11 contains an image of features extracted using GFCC.
https://github.com/mvansegbroeck/featxtra/blob/master/feat/feature-gtf.cc This link shows the method used to evaluate GFCC features. This is a module which is integrated with kaldi toolkit used for Automatic Speech Recognition (ASR).
Can somebody verify the correctness of the GFCC module? Is it really supposed to give a 64*n matrix?
Relevant answer
Answer
the 64 regards the number of filters when computing the gammatone filterbank, which is commonly 64 or 32. Gammatone filterbanks have a different shape and the filter is in time domain compared with triangular filterbanks for MFCC. When they perform the decorrelation via DCT they kept all coefficients for GFCC but you could have changed to 13. In MFCC they kept 13 out of 23 filters (23 filters are the common used for triangular filterbanks). If you check nnet3 implementations in Kaldi you will see that they change the setup for MFCC by making 40 triangular filterbanks and keeping them all, hence 40 "high resolution" MFCC coefficients.
  • asked a question related to ASR
Question
4 answers
Hi Researchers,
I am working on speech domain and I am looking for a "Medical Speech Database" for making an ASR which contain medical diseases and medicine as utterance in speech.
eg: Endoscopy, Asthma are diseases so its purpose, procedure and risk should be present in speech sentence.
Please suggest me some useful link.
Thanks in advance !!!
Relevant answer
Answer
I would recommend these two websites for datasets
  • asked a question related to ASR
Question
3 answers
I am an English trainer for a call center in Vietnam in search of a scalable ASR tool to use in pronunciation courses. The ideal software would provide graded feedback to me and the user. I know of one called Speechace ( https://www.speechace.com/ ), but it only currently works with segmentals of speech, not suprasegmentals. It seems like a helpful tool as far as it goes, returning phonemes, giving a score for utterances, and it even links with Moodle. However, I am wondering if there is something on the market that can also deal with suprasegmentals.
Relevant answer
Answer
I recommend PRAAT software, also there's Speech Analyzer program that allows comparison of native vs. non-native speech. You may check my paper. Both are open access and easy to manipulate.
  • asked a question related to ASR
Question
5 answers
Hi,
I am an English trainer at a call center in Vietnam and am looking for an ASR program that can assist me in teaching pronunciation. I would likely use it as a supplement for out of course work, but would also be interested in something that could be used for testing as well. Something easily scalable would be an extra bonus. Thanks in advance for any suggestions.
Amory
Relevant answer
Answer
  • asked a question related to ASR
Question
3 answers
I have a lot of new cases of colorectal cancers between 2005 and 2010. I calculated age-specific incidence rate for each year (from 2005 to 2010). Now, I want to calculate the incidence rate for all then years which should be just one value.
How can I calculate it?
Relevant answer
Answer
Dear Mehdi
To calculate average annual incidence rate (single value for the ten years), there is a simple method
1. You add all cases registered in the ten years and obtain their average
Summation of cases in 10 years
Average number of cases =---------------------------------------------------------= A
10
2. Obtain the mid-period population by either similar method to 1 above or
just take the average of the population in the middle two years ( population of 2007 +population of 2008) divided by two = B
A
The average annual incidence rate =--------------- X100000
B
With best regards
  • asked a question related to ASR
Question
6 answers
In my data there are five subgroup of age ("0", "1-4", "5-9", "10-14", "15-19"). In some papers, the incidence was calculated for "0-14" and "0-19" using ASRs.
How can I calculate this ASR? (I attached two pages of the article)
Relevant answer
Answer
Recinda Loraine Sherman Thank you. I have seen articles in which ASR was reported per million.
  • asked a question related to ASR
Question
2 answers
I am currently looking for speaker adaption algorithm which captures speaker-dependent information in training data to improve accuracy in ASR. However, most paper I found seem to require prior awareness of speaker identity, which is not included in my data. Is there a speaker adaption algorithm that generates speaker-dependent features without knowing the speakers?
Relevant answer
Answer
You may look at the paper "personalise neonatal seizure detection"
  • asked a question related to ASR
Question
7 answers
For my bachelor thesis I am working on a speech assistant that can be used in bank consulting sessions. For example, the consultant can ask the assistant to provide the history of a specific share's price in a chart.
The focus of my thesis lies on improving the speech-to-text results that we receive from our ASR engine. For this task I need a german corpus from the field of bank consulting conversations.
Since to my knowledge such a corpus does not exist, I need to create my own. As I have no experience in creating corpora, do any of you have suggestions as where to start, what tools to use or how to proceed in general? I was thinking that maybe there are transcriptions of television shows debating share prices or interviews in newspapers that could be used?
Thanks in advance!
Relevant answer
Answer
I'm not personally familiar with any web tools for creating corpora. Perhaps someone else here can help you.
  • asked a question related to ASR
Question
5 answers
Glass use as an aggregate in concrete cause deleterious alkali-silica reaction (ASR) between the cement paste and the glass aggregate which lead to strength loss caused by the weakened interfacial zone.
What is the most effective methods to reduce or eliminate ASR ?
Relevant answer
Answer
  • asked a question related to ASR
  • asked a question related to ASR
Question
3 answers
I am trying to find an standard/guideline for testing a concrete/mortar mixture containing more than one source of aggregate (we have a fine and coarse aggregate). I am already familiar with ASTM C1260, ASTM C1567, ASTM C1293 and AASHTO TP110. But they discuss testing only for one source of aggregate.
We already know that none of them is completely none-reactive (Separate testing ASTM C1260).
Was wondering if anyone knows a standardized test procedure or a guideline for testing the combined aggregate?
Relevant answer
Answer
Hello,
You can take a look at the rilem recommandations hereafter especially the methods aar3 and aar4
You will find some tests dedicated to the aggregate combination qualication. In France these kind of test on concrete is used as a performance concrete test (qualification of the concrete formula expected on site accordant to nf p 18-454).
Best wishes
Maxime
  • asked a question related to ASR
Question
1 answer
Hello,
I am currently modeling Alkali Silica Reaction (ASR) in reinforced concrete structures using Abaqus (FEM). I wrote a subroutine to take into consideration the relationship between stress-state and anisotropic ASR expansion as proposed by Gautam (2016). Basically, it distributed the stresses in the three principal directions. The expansion in the compressed direction is reduced while it is increased where tension (or smaller compressive stresses) governs.
The model works very well for uniform loads and stress-states, but it always crashes when modeling complex structures. After running many analyses, I think I found the error. When using CDP, the stresses fluctuate after the expansion and the non-linear behaviour begin. So I tried to limit this stress variation used in the subroutine by calculating an "effective" stress that can vary a specific amount (0.2MPa for example) in between each step. It made my analysis go further but I'm not even close to the end (which I was able to do assuming isotropic behaviour).
Therefore, I'd like to know if anyone has experience in modeling this type of problem and could give me a few suggestions on how to solve these issues.
Best Regards,
Rodrigo
Relevant answer
Answer
Hey Rodrigo,
Were you able to write the USDFLD? Could you please point me to some resources where I can understand how it works?
  • asked a question related to ASR
Question
2 answers
Dear Colleague,
I am working on a forensic engineering project and from the observation of the problem on the site I thought it is ASR problem. Therefore, I performed XRF and I am trying to know the potassium percentage on the XRF results that could give me idea if it is ASR or not?
Relevant answer
Answer
First, what XRF system are you using? The calibrations may have a bias one way or another (full disclosure, I used to work for Bruker)
Second - why not look for Na? It seems to be a stronger predictor of ASR
Third, what may be more important than bulk Na or K is their ratio relative to one another based on this research:
  • asked a question related to ASR
Question
2 answers
I am working on a journal article which has been finished now working on English editing and I do feel that a contribution from a native speaker in this field of study will boost the paper quality and can be published in Q1 journal, if anyone willing to collaborate please let me know I will take it to my supervisor, thank you.  
Relevant answer
Answer
I can help you. Can mail me for further details.
  • asked a question related to ASR
Question
3 answers
I want to develop an Automatic Speech Recognition(ASR) for Under-Resourced Language : Konakni (Regional Language spoken in Konkan part of India). Since there is no standard Speech Database is  available, I want to develop it from scratch.
I need sound knowledge of following parameters:
  1. Hardware requirement
  2. minimum bit rate
  3. Recording environment
  4. No of Hours training data
  5. no of speakers (Male,female, children,demographics of speakers) etc.
Waiting for reply.
Relevant answer
Answer
You can start with Kaldi. It has multiple recipes for building different types of ASR from scratch. 
  • asked a question related to ASR
Question
3 answers
I am in a site where I cannot economically obtain other suitable aggregates for making concrete.
Relevant answer
Answer
additional to Dr Prunell: Coarse aggregate grading can be preffered coarser to decrease the expansion levels in long term as seen the results of study given below. The method can also be decreased paste volume necessary with decrease of total surface area according to similar consistency and get economy.  However the fresh mix can be tend to harsh mix, so it must be checked
0.075-5 mm size portion of aggregates that more tends to ASR may be decreased and 0-0.075 mm can be increased . Because reactive aggregates generally works as pozolanic materials below 0.075 mm in long term. So there is no need extra cementitious materials such as silica fume etc.
  • asked a question related to ASR
Question
15 answers
Hello,
I am looking for information on ASR concrete tests on samples subjected to real climate impacts.
I have found information on tests that measure expansion of the specimens using the DEMEC strain gauge.
I am interested in other measurement methods (optical fibers, strain gauges attached inside the sample) and current international research programs in this field.
Regards,
WojtekK
  • asked a question related to ASR
Question
5 answers
OM-SEM analysis on a 10-year old concrete pavement in a bus stop have shown serious damage due to AAR. It turned out that the aggregates responsible for the formation of expansive products (both amorphous gel and microcrystalline phases) and the damage are serpentinite/greenstone/greenshist. There is clear evidence for reaction between the pore solution and some mineral phases (e.g. Ca-Mg amphibole, serpentinite, etc.). The reactions, however, are not fully understood so far, especially because some of the observed mineral phases are being stabile in high pH environment. Has anybody seen serpentinite, greenstone undergo to AAR (ASR)? Thanks!
Relevant answer
Answer
Dear Mr. Pinter:
Please find attached a flow sheet demonstrating the supergene alteration of metabasic and metaultrabasic rocks to be published in Earth Science Reviews (2016) online and entitled:
Residual clay deposits on basement rocks: The impact of climate and the geological setting on supergene argillitization in the Bohemian Massif (Central Europe) and across the globe.
It may help you get a bit closer to a solution.
Viel Erfolg
Harald G. Dill
  • asked a question related to ASR
Question
4 answers
how introducing phoneme length parameter into standard HMM model and producing new one to be used in building accurate ASR system using HTK? thanks for any reply
Relevant answer
Answer
Thanks to all of you, these replies are so interesting and helpful.
  • asked a question related to ASR
Question
4 answers
ASR = Rp x area/2. The area refers to which area? Sample area or current collector area?
Relevant answer
Answer
It depends on your experiment. If you are studying electrochemical performance of electrolytes, I mean you are using a symmetric cell configuration and you painted the electrodes on both side of electrolyte you have to consider for the ASR the electrolyte area. If you are studying the entire cell for Area you need the current collector area. It depends on your configuration!!!
  • asked a question related to ASR
Question
1 answer
CRT glass is manufactured at very high temperature (above 1000 deg.). Therefore, CRT glass should be "Crystalline" in form which is mostly non-reactive. However, some of the researchers (like the following reference) reported alkali–silica reaction (ASR) while using CRT glass as fine aggregate in cement mortar. As CRT glass is in non-reactive form, how ASR problem occur in such mortar.
Ling, T.-C., Poon, C.-S. (2013) Effects of particle size of treated CRT funnel glass on properties of cement mortar. Materials and Structures; 46(1-2): 25-34.
Relevant answer
Answer
Dear Amrul:
Cement ( 62-76% СаО, 20-23% SiO2, 4-7% А12О3, 2-5% Fе2О3, 1-5% MgO). 62-76% СаО, 20-23% SiO2, 4-7% А12О3, 2-5% Fе2О3, 1-5% MgO;  Glass (Na2O  PbO * 6SiO2).Cement and glass contain basic (Na2O) , acidic (SiO2) , amphoteric oxide (Аl2О3 ). When heating the basic oxide reacts with acidic oxide. The oxides are dissolved in the water. In the water they also react. The CRT glass is crushed into small particles. The crushed material is better able to interact chemically with another substance. Form strong covalent bonds cement mortar
Best Regards.
  • asked a question related to ASR
Question
3 answers
I want to know availability of fully developed ASR systems for any Indian language - either LVCSR systems or limited vocabulary systems deployed in enquiry based services. (i) Is anyone available for purchase on a licence basis ? (ii) Even if it is not available for buying, has any system been practically deployed for some application and can be tested by public through that service?
Relevant answer
Answer
I thought I made it very clear: Fully developed ASR for any Indian language.
I must also mention that ASR neither understands any language nor actually recognizes the content of speech. Today's best ASR is just a transcription system, which matches an input stream of speech to an output stream of words from its pre-stored vocabulary. Understanding the speech ? Forget it. Nobody is even thinking of it, even at the design stage.
  • asked a question related to ASR
Question
2 answers
bottleneck features VS neural netowrk
Relevant answer
Answer
Thanks so much
  • asked a question related to ASR
Question
4 answers
I am thinking of using waste crushed glass powder due to its high silica content to stabilize expansive soils with the addition of some additives to take care of possible ASR.
Relevant answer
Answer
Hi, if we exclude the cement addition, will the glass powder start a pozolanic reaction upon the contact with expanding clay minerals? I don't think so. The addition of glass powder might increase the shear strength due to enhanced frictional resistance. 
If you want to be a friend of the environment, minimize or don't use the cement as a stabilizing agent. Lime or ash were known for their performance of stabilizing expanding soils.
Anyways, good luck with your research...
  • asked a question related to ASR
Question
6 answers
A segmentation process gives phonetic segments of variable length, then this segments are labeled one out of many phonemes(classification). Just i would want to know how this (segmentation+classification) task is  different from recognition process.
Relevant answer
"Acoustic decoding" ( aka "acoustic recognition") is precisely this segmentation + classification process. But ASR (automatic speech recognition) at present involves much more than this. Which amounts, more or less, to seeing whether the labelled segments you obtained make sense within a particular language (or rather a sample of a particular language).
Hope this helps.
  • asked a question related to ASR
Question
1 answer
I want to mix two LM, one in word and another sub-lexical item (e.g. a syllable).
Relevant answer
Answer
It ultimately depends on what you want to do...
SRILM has a tool for model interpolation. But this usually implies having a common symbol set shared by the models.
The following paper describes the use of hybrid models (mixing different classes of symbol in the same model), as applied to OOV detection:
L. Qin and A. Rudnicky OOV Word Detection using Hybrid Models with Mixed Types of Fragments. Proceedings of Interspeech, 2012, Portland, OR US.
Finally, you can decode the speech multiple times, each time using a different model. Of course you then have to figure out alignment and possible fusion, if that's relevant.
It really does depend on what you're after. Perhaps you can be more specific about your goal.
  • asked a question related to ASR
Question
9 answers
I understand lexicon generation can benefit a lot from diacritization. Diacritics in training data will help acoustic modeling. The question is, if diacritics in language model improves the performance of Arabic speech recognition? It doesn't help in my experiment, I would like to know if other people have similar experiences. Thank you.
Relevant answer
Answer
I do not know the diacritics issues in ASR. However in Arabic Text-To-Speech, computing the correct diacritics in a written sentence where diacritics are missing is also a challenge.
According to my experience in TTS, diacritics in standard Arabic are mainly related to two kinds of ambiguities: grammatical (or case ending suffix) and lexical. Case ending suffix may be hidden by a silent diacritics or an elision. Lexical diacritics related ambiguities occur mainly in triliteral words (Verb/Noun ambiguities, kataba/ kutub) with or without agglutinated pronouns, personal/possessive .
In order to measure in a model a significant improvements, one should measure separately the improvement for lexical ambiguities related the triliteral words with/without diacritics in training data for the acoustic model. Or, one could measure the improvements according to a predefined typology related to diacritics in ASR acoustic model.
Cheers
  • asked a question related to ASR
Question
4 answers
In terms of practically useful applications, do you know significant applications for a system that can recognize speech that contains words from more than one language?
Relevant answer
Answer
Dear Sir, good evening. I have relatively shallower understanding of the signal than those trained in signal processing but have good observation of signal shapes while manual sound annotation. On that basis I can say that distinct phones can be identified relatively easier with signal shape and spectrum. But I have also some experience that one word listened 50 times is perceived at least 4 ways from the same listener (myself). There is confusion even in number of the phones. Those cases seem more difficult to me for teaching to machine.
  • asked a question related to ASR
Question
14 answers
Is there any speech recognition system with real-time recognition capability? Not pocketspinx or sphinx. Some newer ones that can perform well on real tasks please.
Relevant answer
Answer
I suggest using ATK the real-time API for HTK (Hidden Markov Model toolkit) of University of Cambridge. Here is the link: