At Pneuma Solutions, we stand by our commitment to excellence. So does every other software developer, right? But the difference between us and everyone else is that Ours is a cause, not just the bottom line. The day we become irrelevant will be the day we can say accessibility has become so ubiquitous as to be commonplace, and that will be a day worth celebrating.
Until then, we work in a pool of companies following distinct strategies in an attempt to make accessibility universal. We each have our competing vision of what that looks like, and no one vision is better than or more correct than the other. There is more than enough work to go around. So while we compete, we are still on the same team with the same goal of making the World a more accessible place. In the end, more options means greater emphasis on accessibility in the marketplace and more tools available to us as consumers.
In that spirit of positive sportsmanship, we present a comparison of our flagship product, Scribe, versus RoboBraille. It’s a comparison a few people in the blindness community have drawn, and rather than shy away from the whispers, we’re going to examine both products head on and let you be the judge of which product works best for your needs. We aren’t jealous. We don’t mind sharing space in your toolbox with as many accessibility tools as you need to experience your digital World.
Overview
RoboBraille and SensusAccess are two closely related products provided by the Danish company, Sensus ApS. They are both based on the same conversion engine. RoboBraille is offered free of charge for individual non-commercial use, while SensusAccess is marketed to educational institutions with a free demo.
In this comparison, the name RoboBraille will be used when referring to features common to both RoboBraille and SensusAccess.
Input formats
Both Scribe and RoboBraille support a variety of input formats.
|
|
Scribe |
RoboBraille |
|
|
Yes |
Yes |
|
Microsoft Office |
.docx .doc .rtf .xlsx .xls .pptx .ppt |
.docx .doc .rtf .pptx .ppt |
|
Hyper Text Language |
N/A |
XML and HTML |
|
Open Document Format |
.odt .ods .odp |
N/A |
|
eBooks |
.epub .mobi |
.epub .mobi |
|
Images |
.bmp .dcx .djvu, .djv .gif .jpeg, .jpg .j2k, .jp2, .jpx .pcx .png .tiff, .tif |
.bmp .dcx .djv .gif .jpeg, .jpg .j2k, .jp2, .jpx .pcx .png .tiff, .tif |
|
Plain Text |
.txt) |
.txt, .asc |
|
LaTex |
N/A |
LaTeX (.tex) |
The imbalance is intentional. We deliberately chose not to support the few input formats RoboBraille supports that we do not. In our testing, RoboBraille’s conversion of a simple LaTeX document, for example, was not of a sufficiently good standard. LaTeX is usually used for documents that are heavy on scientific and mathematical notation, and remediating such documents is difficult. We plan to take on this challenge, but we will not incorporate until we are satisfied it is acceptable.
As for raw HTML or XML input, Scribe is designed for converting stand-alone, downloadable documents, not web pages in all of their hyperlinked, interactive complexity. An HTML file is often not useful on its own; for example, it may refer to images or style sheets. And we wouldn’t be able to meaningfully translate interactive features of a web page to all output formats. That’s why we choose not to provide support for HTML or XML input.
Consistency of output formats
RoboBraille doesn’t offer the same output format options for every input format. For example:
- When using LaTeX (.tex) input, the only supported output option is HTML.
- When converting a PowerPoint slide deck, the only supported outputs are PDF, HTML, and RTF.
- RoboBraille supports DAISY output for some input formats, such as .docx and .doc, but not for others, such as PDF.
By contrast, Scribe supports any combination of input and output format, so print-impaired users can easily get the information in whatever format works best for them, regardless of the original format.
Multilingual support
RoboBraille appears to have some limited automatic language detection, but in our testing, it wasn’t consistent or reliable. When converting an all-English PDF document to EPUB with media overlay, some of the speech output was incorrectly rendered in a non-English language. On the other hand, when providing a multilingual PDF document as input, there was no multilingual option when requesting audio (MP3) output.
By contrast, Scribe automatically detects the language or languages of the input document by default. This allows output formats such as tagged PDF, HTML, and EPUB to be tagged with the detected language or languages. For multilingual documents, this detection and tagging is done on a per-paragraph basis. This also means that when converting to audio, either in a stand-alone MP3 file or as part of a DAISY ebook, Scribe can automatically switch to an appropriate text-to-speech voice for each language in a multilingual document.
AI-powered image descriptions
Scribe can automatically add descriptions to images that aren’t already tagged with alternate text, using a cloud-based, AI-powered image description service. These image descriptions are included in all supported output formats, including audio and Braille. Based on our testing, RoboBraille doesn’t appear to have this feature.
Audio output
RoboBraille currently gives the user greater control over the audio output, though only when converting to a stand-alone MP3 file. For some output languages, it provides multiple dialects and voices. However, the level of flexibility is inconsistent; for example, only one American English voice is supported, while multiple voices are supported for more obscure languages. For EPUB with media overlay, RoboBraille provides no control over the language, voice, or speaking rate. When converting to DAISY, RoboBraille lets the user choose a language, but not a specific dialect or voice, and doesn’t provide control over the speaking rate.
In Scribe, the language selection for the audio output is determined by the input language or languages. Scribe currently only allows the user to specify the gender, not the dialect or a specific voice. We plan to add more control over the voice selection. However, the settings that we do have, including speaking rate, are consistently available for both stand-alone MP3 and DAISY output.
Scribe’s text-to-speech output is based on the most cutting-edge, natural-sounding text-to-speech technology on the market. RoboBraille’s text-to-speech output is much more monotonous, at least in English. Scribe also integrates more advanced text pre-processing, to effectively handle input characteristics such as mixed capitalization and repeated decorative punctuation. Finally, while RoboBraille converts a document to plain text before converting that text to audio, Scribe intelligently produces speech output directly from rich text. This means, for example, that Scribe’s text-to-speech conversion process knows where the images are in a document, and can add a sound effect to mark each image, played at the same time as the image description is being spoken.
Ebook output
RoboBraille supports conversion to DAISY, EPUB, EPUB with media overlay, and MOBI ebook formats. However, as noted above, the set of available output formats isn’t consistent for all input formats. Scribe currently supports DAISY (text and audio), EPUB (text only), and MOBI. We plan to add support for EPUB with media overlay, and as with all Scribe output formats, these will be consistently available for all input formats.
Braille output
RoboBraille’s Braille output support is more granular than Scribe’s. RoboBraille has separate drop-down menus for choosing the Braille language (e.g. Unified English Braille or U.S. English) and the level of contraction (e.g. grade 1 or grade 2). Scribe currently offers a single, combined list of preset Braille translation options. RoboBraille also supports output to several variations of the Portable Embosser Format (PEF), whereas Scribe currently only supports output to legacy .brf files.
Tagged PDF output
RoboBraille gives the user more flexibility when creating a tagged PDF file; the user can choose between “text over images” and “images over text”. However, this choice is forced, and it requires the user to understand the tradeoffs. By contrast, Scribe keeps it simple for the user by intelligently choosing the best option based on the input document.
Large print support
In our testing, RoboBraille only offered large print as an option when converting to EPUB. By contrast, Scribe offers large print for EPUB, PDF, Microsoft Word (.docx), and online reading.
Other output formats
In addition to the output formats discussed above, both RoboBraille and Scribe offer conversion to Microsoft Word (.docx) format.
RoboBraille offers conversion to Excel in some cases. We will investigate conversion to an Excel spreadsheet, but we only want to offer that option for input documents where that conversion makes sense, specifically documents that are primarily tabular data.
Both RoboBraille and Scribe offer conversion to a downloadable HTML file. Scribe’s downloadable HTML output also embeds any images from the original document.
RoboBraille offers conversion to RTF, whereas we currently only support the more full-featured and efficient .docx format.
Finally, RoboBraille offers conversion to plain text (.txt) format, whereas Scribe doesn’t support that option, because plain text is a primitive format that lacks crucial features such as reliable Unicode support (crucial for non-English languages) and navigable document structure. We believe that given the variety of other formats we offer, which are supported by all types of software and devices, .txt output isn’t necessary.
Online reading and preview
Scribe’s conversion process is much more interactive than RoboBraille’s. Scribe lets a user preview a converted document before downloading their desired conversion format(s). Scribe also lets the user read the full document as a web page on any device, without having to download anything at all. Scribe makes it easy and efficient to do multiple conversions from the same source document, whereas RoboBraille requires the user to go through the conversion process from scratch each time. Finally, while Scribe supports email notifications as RoboBraille does, for most output formats, Scribe also offers real-time progress updates and lets the user view or download the result without leaving the web page.
Integrations
At the time of this writing SensusAccess has more flexible integration options than Scribe. SensusAccess provides ready-to-use integrations with learning management systems (LMS), including Moodle, Canvas, Blackboard Learn, Sakai, and Desire2Learn. These integrations are provided through an add-on service called SensusAccess LTI. SensusAccess also provides a web API that enables development of custom integrations.
Scribe offers a JavaScript widget for embedding the Scribe conversion workflow into customers’ websites. Scribe does not yet offer LMS integrations or a customer-facing API.
Reliability and infrastructure
In our testing of the free service, RoboBraille didn’t reliably convert documents to the requested output format. For example, attempts to convert a PowerPoint slide deck into PDF, HTML, and RTF all failed, though we did get an unrequested conversion to plain text (.txt). Similarly, an attempt to convert a Word document (.docx) into tagged PDF resulted in a conversion to plain text instead. Finally, in an attempt to convert a PDF document into HTML, we got two different conversions to plain text, from two different versions of the RoboBraille server software.
RoboBraille seems to use a patchwork of conversion processes for different combinations of input and output formats. The patchwork nature of the software, including version numbers and back-end email addresses, is exposed to the end-user in the conversion result emails. This software appears to be deployed on a handful of dedicated servers running the legacy .NET Framework. By contrast, Scribe is powered by a unified conversion engine deployed on a highly secure modern, scalable, AWS-based cloud infrastructure, with a simple user experience that’s not needlessly cluttered with implementation details.
Conclusion
Some features, such as audio output, are purely subjective. Others may rise to the level of must have. In that case, we’ve done our best to give an objective comparison as of this writing. Only you can decide what is essential.
We hope the information has proven useful in helping you make an informed choice. We commend other companies for contributing to the general knowledge pool. We believe our product is a considerable competitive option, we have worked hard to develop a sophisticated platform and may be a little biased.
Do you have any questions? We’d love to spend a little time answering them. Get in touch directly or leave us a comment.
コメントを残す