Citation Link: https://doi.org/10.25819/ubsi/10225
The ‘Conditional Voice Recorder’: Data practices in the co-operative advancement and implementation of data-collection technology
Source Type
Book
Series
Working paper series / SFB 1187 Medien der Kooperation
Subjects
Smart Speaker
Voice Assistant
Intelligent Personal Assistant
Voice-user interface
Human-Machine-Interaction
Data Practices
Research Practices
Hot-word detection
Data collection
DDC
302.23 Medien (Kommunikationsmittel), Medienwissenschaft
GHBS-Clases
Issue Date
2022
Abstract
Stationary voice-controlled systems are installed in an increasing number of households. The devices are operated primarily via voice-user interfaces, which evaluate the spoken commands cloud-based, and are aligned to the principles of interpersonal interaction. This raises questions about the integration of the devices into everyday practices carried out in the household: How is use of Smart Speakers negotiated situationally, embedded in interpersonal interactions, and (how) are aspects of data privacy, data processing and potential exploitation reflected by the users? The project "Un/desired Observation in Interaction: Intelligent Personal Assistants" in the CRC "Media of Cooperation" approaches these questions empirically.
However, such an investigation of smart speakers faces the methodological challenge that this requires voice data documenting not only the use of the smart speaker itself, but also the contexts of the use that go beyond mere "voice commands". Therefore, a "Conditional Voice Recorder" (CVR), a technology developed in Nottingham by Porcheron and colleagues (2018), was brought to bear to create audio recordings of usage contexts. These include not only the voice command itself, but also a few minutes before and after the smart speaker is addressed. However, the original device required further technical development to be compatible not only with Amazon's smart speaker model, but also products from other providers (Google and Apple).
The paper reflects on this advancement and the implementation of the CVR – i.e. our own research practices – as data practices. On the one hand, it makes visible which (otherwise opaque) data were collected and processed during the advancement, how the usage of the CVR itself is inscribed in the data recorded with it, and which data practices were carried out in the evaluation. On the other hand, it documents the advancement and application of the CVR to enable other studies with it (or similar technologies).
However, such an investigation of smart speakers faces the methodological challenge that this requires voice data documenting not only the use of the smart speaker itself, but also the contexts of the use that go beyond mere "voice commands". Therefore, a "Conditional Voice Recorder" (CVR), a technology developed in Nottingham by Porcheron and colleagues (2018), was brought to bear to create audio recordings of usage contexts. These include not only the voice command itself, but also a few minutes before and after the smart speaker is addressed. However, the original device required further technical development to be compatible not only with Amazon's smart speaker model, but also products from other providers (Google and Apple).
The paper reflects on this advancement and the implementation of the CVR – i.e. our own research practices – as data practices. On the one hand, it makes visible which (otherwise opaque) data were collected and processed during the advancement, how the usage of the CVR itself is inscribed in the data recorded with it, and which data practices were carried out in the evaluation. On the other hand, it documents the advancement and application of the CVR to enable other studies with it (or similar technologies).
File(s)![Thumbnail Image]()
Loading...
Name
WPS_23_The_Conditional_Voice_Recorder_2.pdf
Size
1.23 MB
Format
Adobe PDF
Checksum
(MD5):36aea8d9b9650ed1f9a11d091ee3b98e
Owning collection