Re: [AUDITORY] online data collection for auditory experiments

["Barbour, Dennis" <dbarbour@xxxxxxxxx>]

A few people have approached me about the availability of our online hearing tests in order to help maintain social distancing in their clinics. We have a beta version of the machine learning audiogram accessible by a web browser. It is the version used for this paper

https://journals.lww.com/ear-hearing/Abstract/2019/07000/Online_Machine_Learning_Audiometry.14.aspx

and also implements the bilateral audiogram described here

https://www.ingentaconnect.com/content/dav/aaua/2018/00000104/00000005/art00010

This audiometer was the first to use WebAudio and may still be the only one. It should work on any device that can run Chrome, though we have not thoroughly tested that.

The beta was intended to demonstrate the feasibility of running active machine learning code in the cloud while any client device delivers sounds locally. A clinician or experimenter was expected to be operating the audiometer, but it can be operated remotely, though there would not be an easy way to calibrate in that case.

If this is a tool that might help you, I can give you an account. Please get in touch.

Dennis Barbour

PS A standard Hughson-Westlake procedure is also implemented.

On Apr 15, 2020, at 2:17 PM, Christer P. Volk <cvo@xxxxxxxx> wrote:

Hi Arnaud and all

 

I would also be interested in hearing how people are doing level calibration in online tests and I can share what we have tried in my team.

 

For informal playback (of road noise auralisation), we simply have a recording of a person talking (at a normal vocal effort) and ask subjects to adjust their volume to a realistic level of a person standing 1 meter away. Then everything else is adjusted relative to that. We have tested this in our lab with 24 subjects and found that to work relatively well (mean corresponding to 64 dBA), but between the subject with the lowest SPL and the subject with the highest SPL (on the same physical setup) you might still get a 15 dB difference. In this case we used self-reported normal hearing subjects and suspect that part of the difference might actually be explained by differences in hearing.

 

Of more concern for our purpose with online testing, is the range of frequency responses of the subjects’ headphones and their background noise levels, affecting both what is heard and the subjects’ level of concentration. I would love to hear what solutions have been tried by others!

 

We also have a panel of trained subjects. A small proportion of that training is home training. For that purpose we have supplied all with the same closed-back headphone model and the same sound card. This allows for OK experiment control, although operating system settings are a challenge and handling background noise is still not solved. This solution might work if your subjects are students that can be trusted to return the equipment (or ship it to the next subject).

 

In terms of testing involving subjects with hearing loss, I would not trust subjects above a certain age to set up and run everything correctly. Just following a guide to disable Windows sound enhancements could be a challenge!

 

Great to see how everybody is helping each other here in this special situation!

Best regards

Christer P. Volk 
Ph.D. 

Specialist 
SenseLab, FORCE Technology 

From: AUDITORY - Research in Auditory Perception <AUDITORY@xxxxxxxxxxxxxxx> On Behalf Of arnaud norena
Sent: 15. april 2020 09:47
To: AUDITORY@xxxxxxxxxxxxxxx
Subject: Re: online data collection for auditory experiments

 

Hi all,

 

we are very much interested in doing some online experiments but we are wondering how we can have a sense of the sound level that is deliverd for the tested subject. Do you simply ask the subjects to set the level at a comfortable loudness? How do you manage if subjects have hearing loss? Do you test for hearing thresholds? etc.

 

Any feed-backs on this topic would be very useful!

 

Thanks!

arnaud norena 

 

Le mer. 15 avr. 2020 à 09:38, Picinali, Lorenzo <l.picinali@xxxxxxxxxxxxxx> a écrit :

Thanks for the list indeed!

 

In case you want to do some dynamic binaural spatialisation online, we have created a _javascript_ wrapper of our 3D Tune-In Toolkit (https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0211899), which can be found here:

 

https://github.com/3DTune-In/3dti_AudioToolkit_JavaScript

 

And here is a working example, which is currently hosted on Heroku:

 

http://online-toolkit.3d-tune-in.eu/

 

We have also created a web app allowing users to import their own audio files (only MP3 is supported), create soundscapes and interact with them (this is also hosted on Heroku). You can find the app here:

 

https://pluggy-plugsonic.herokuapp.com/

 

and here is a brief conference paper describing its functionalities, together with the functionalities of a web-based audio editor created with the Web Audio API (plugsonic.pluggy.eu/sample/):

 

https://secure.aes.org/forum/pubs/conferences/?elib=20435

 

I hope this helps!
Lorenzo

 

-- 
Dr Lorenzo Picinali
Senior Lecturer in Audio Experience Design
Director of Undergraduate Studies
Dyson School of Design Engineering
Imperial College London
Dyson Building
Imperial College Road
South Kensington, SW7 2DB, London 
T: 0044 (0)20 7594 8158
E: l.picinali@xxxxxxxxxxxxxx

http://www.imperial.ac.uk/people/l.picinali

www.imperial.ac.uk/design-engineering-school

---

From: AUDITORY - Research in Auditory Perception <AUDITORY@xxxxxxxxxxxxxxx> on behalf of Etienne Gaudrain <egaudrain.cam@xxxxxxxxx>
Sent: 14 April 2020 10:35
To: AUDITORY@xxxxxxxxxxxxxxx <AUDITORY@xxxxxxxxxxxxxxx>
Subject: Re: [AUDITORY] online data collection for auditory experiments

 

This email from egaudrain.cam@xxxxxxxxx originates from outside Imperial. Do not click on links and attachments unless you recognise the sender. If you trust the sender, add them to your safe senders list to disable email stamping for this address.

 

Thank you Felix, this is a great list!

It's never very clear what the audio capabilities are for each of these tools, but it looks like they can all at least play a sound (and sometimes generate tones and the likes). Manipulating the sounds is another story.

Some of our experiments are adaptive and need generation of speech stimuli on the go. For that we've written a Python server that can do some of these manipulations (like changing voice pitch, or vocode). This is still work in progress, but what's done is fully functional:

https://github.com/egaudrain/VTServer

The documentation is available here:

https://egaudrain.github.io/VTServer/

You can also try it here:

https://dbsplab.fun/demo/

 

We keep the files you upload on the server 24h before they get deleted... and nobody's going to listen to them.

So far it's running on a pretty basic virtual private server that costs about 8€/month. We'll see with usage if we need to scale to something more powerful, or more elastic.

If you have comments, or if you want to contribute, don't hesitate to fork it or send pull requests. So far I've written a very rudimentary install script for Ubuntu 18.04 LTS to install the thing as a systemd service, but if someone has knowledge in, e.g. PyPi/pip packaging, it could make installation much easier.

Cheers,
-Etienne

-- 
Etienne Gaudrain, PhD

Centre de Recherche en Neurosciences de Lyon - CNRS UMR 5292
Université Lyon 1
50 av. Tony Garnier
69366 Lyon Cedex 7, FR

 

UMCG, Afdeling KNO, BB20
PO Box 30.001
9700 RB Groningen, NL

Phone +31 5036 13290
Skype egaudrain

Note: emails to this address are limited to 10 MB. To send larger files, use egaudrain.cam@xxxxxxxxx.

 

 

On Thu, 9 Apr 2020 at 11:59, Felix Dobrowohl <F.Dobrowohl@xxxxxxxxxxxxxxxxxxxx> wrote:

Hi Jessica and List,

 

We’ve been compiling a resource of online experiment platforms for behavioural auditory research which might be of use for some of you.

Note, experiment builders which are listed as hosting by Pavlovia can potentially be hosted elsewhere.

 

Have a good long weekend everyone,

 

Cheers,

Felix

 

Integrated Service providers 

Name	GUI	hosting	Cost	Comment	URL	Tutorials/Templates
Qualtrics	yes	provided	~2500$/year	Survey builder with integrated audio/video support. Limitations for randomisation. For surveys only.	Qualtrics.com	Qualtrics Tutorials
Gorilla.sc	yes	provided	1.08 USD per participant	Online experiment builder with drag & drop GUI but limitations for randomisation among others.	Gorilla.sc	gorilla.sc tutorials
Testable	Yes	provided	Mixed payment. Potentially free	Very simple experiment builder for basic behavioural experiments. Automatic credit granting via SONA may not be possible.	Testable.org	Testable tutorials
Inquisit Web	No	provided	200 USD/month	Scriptable experiment platform, which requires a client installation from the participant, but in turn provides the potentially best control and precision	Inquisit Web Link	Inquisit templates
Labvanced	Yes	provided	1080 EUR/quarter	Versatile interface and large amount of tutorials and support online	Labvanced.com	Labvanced Video tutorials

 
Experiment builder software (free) 

Name	GUI	Coding language	hosting	Comment	URL	Tutorials,  Templates
PsychoPy3	Yes	Python	Pavlovia	Similar in structure to e-prime, pavlovia implemented	PsychoPy.org	Link
PsychoJS	No, works in conjunction with PsychoPy	_javascript_	Pavlovia	_javascript_ library for PsychoPy experiments	psychoJS	Link
JsPsych	No	_javascript_	Pavlovia	_javascript_ library	JSPsych.org	Link
PsyToolkit	No	own	provided	Basic script-based experiment builder with its own scripting language	PsyToolkit.org	Link
NodeGame	No	_javascript_	self	_javascript_ based experiment builder with option for multi-participant designs	NodeGame.org	Link
Tatool	Yes	_javascript_	self	Simple experiment builder, supporting both a GUI or _javascript_.	Tatool-Web.com	Link
LionessLab	Yes		self	A simple online experiment builder that allows for interactive experiments	Lioness-Lab.org	Link
O-tree	No	Python	self	Python based experiment builder with option for multi-participant designs	OTree.org	Link
LabJS	Yes	HTML, _javascript_	Pavlovia	User friendly HTML experiment builder with some GUI functionality	LabJS	Link
OpenSesame	Yes	Python, _javascript_	self	Online experiment builder with minimal GUI, using both _javascript_ and python	OpenSesame	Link
R-Shiny	No	R	self	R package to build interactive web apps from R Studio, extendable with CSS, html and _javascript_	Shiny Rstudio	Link

  

Hosting platforms 

Name	Cost	Comment	URL
Pavlovia	1500 pounds/year	Hosting platform by the University of Nottingham, designed for PsychoPy/PsychJS but also hosts LabJS and jsPsych	Pavlovia.org
Amazon Web Services	variable	Web hosting service by Amazon	AWS
Firebase	Storage: 0.03$/GB  Traffic: 0.15$/GB	Web hosting service by Google	firebase.google.com
Heroku	variable		Heroku.com

 

Containers (runtime environments required to run experimental code on servers) 

Name	Cost	Comment	URL
JATOS	Free, open source		JATOS.org
Experiment factory	Free, open source		ExpFactory.github.io

 

 

Felix Dobrowohl | Senior Technical Officer - MARCS Institute

Western Sydney University – Bldg 5.G.57 Bankstown Campus

P: +61 02 9772 6723

E: F.Dobrowohl@xxxxxxxxxxxxxxxxxxxx

 

westernsydney.edu.au

 

From: AUDITORY - Research in Auditory Perception [mailto:AUDITORY@xxxxxxxxxxxxxxx] On Behalf Of Jessica Louise Slater, Ms
Sent: Thursday, 9 April 2020 12:07 AM
To: AUDITORY@xxxxxxxxxxxxxxx
Subject: [AUDITORY] online data collection for auditory experiments

 

Dear list,

 

Like many labs, we are currently evaluating options for running auditory experiments online. I wondered if others would be willing to share their experiences with online data collection, or resources they have found helpful - particularly pertaining to the challenges of auditory stimulus presentation (e.g. timing precision)? We currently have adaptive tasks running in Matlab so are especially interested to hear about experiences using Matlab-based solutions, but any insights very welcome. 

 

Best,

 

Jessica    

 

 

 

-- 

Arnaud Norena 

- DIRECTEUR DE RECHERCHE 2ème classe CNRS 

- FACULTE DES SCIENCES

Aix-Marseille Université - Laboratoire Neurosciences Sensorielles  et Cognitives - ST CHARLES - Case B - 3 Place Victor Hugo - 13003 Marseille

Tél: +33(0)4 13 55 08 65

Site : http://www.univ-amu.fr - Email : arnaud.norena@xxxxxxxxxxx

Afin de respecter l'environnement, merci de n'imprimer cet email que si nécessaire.