Here\u2019s another question to ask about ASR Technology. Why should I, the consumer, look beyond the big three\u2013Google, Apple, Microsoft\u2026make that four, IBM to assist me in meeting all my ASR Requirements? Obviously they have the biggest R&D budgets and attract the best talent so their technology should be the best, right?<\/p>\n
The answer is, it depends.<\/p><\/blockquote>\n[\/vc_column_text][\/vc_column][vc_column column_padding=”no-extra-padding” column_padding_tablet=”inherit” column_padding_phone=”inherit” column_padding_position=”all” column_element_spacing=”default” background_color_opacity=”1″ background_hover_color_opacity=”1″ column_shadow=”none” column_border_radius=”none” column_link_target=”_self” column_position=”default” gradient_direction=”left_to_right” overlay_strength=”0.3″ width=”1\/3″ tablet_width_inherit=”default” tablet_text_alignment=”default” phone_text_alignment=”default” animation_type=”default” bg_image_animation=”none” border_type=”simple” column_border_width=”none” column_border_style=”solid”][image_with_animation image_url=”16005″ image_size=”full” animation_type=”entrance” animation=”Fade In” hover_animation=”none” alignment=”” border_radius=”15px” box_shadow=”small_depth” image_loading=”default” max_width=”125%” max_width_mobile=”default”][\/vc_column][\/vc_row][vc_row type=”full_width_background” full_screen_row_position=”middle” column_margin=”default” equal_height=”yes” content_placement=”middle” column_direction=”default” column_direction_tablet=”default” column_direction_phone=”default” bg_color=”#f7f7f7″ scene_position=”center” top_padding=”5%” constrain_group_1=”yes” bottom_padding=”5%” text_color=”dark” text_align=”left” row_border_radius=”none” row_border_radius_applies=”bg” overflow=”visible” advanced_gradient_angle=”0″ overlay_strength=”0.3″ gradient_direction=”left_to_right” shape_divider_position=”bottom” bg_image_animation=”none” gradient_type=”default” shape_type=””][vc_column column_padding=”no-extra-padding” column_padding_tablet=”inherit” column_padding_phone=”inherit” column_padding_position=”all” column_element_spacing=”default” background_color_opacity=”1″ background_hover_color_opacity=”1″ column_shadow=”none” column_border_radius=”none” column_link_target=”_self” column_position=”default” gradient_direction=”left_to_right” overlay_strength=”0.3″ width=”2\/3″ tablet_width_inherit=”default” tablet_text_alignment=”default” phone_text_alignment=”default” animation_type=”default” bg_image_animation=”none” border_type=”simple” column_border_width=”none” column_border_style=”solid”][vc_column_text]\n
ASR Technology Hits and Misses<\/h2>\n
For example, you want Google to turn on the lights\u2013\u201dGoogle! Turn on the driveway lights.\u201d Or, \u201cSiri! Play my, I\u2019m really depressed mix.\u201d Or \u201cAlexa, I need a Vegan pizza, light pepperoni and cheese.\u201d All of these technologies that use ASR to pick up on voice commands work pretty well.<\/p>\n
However, there are a number of cases where these ASR technologies have challenges. One pretty simple example is when I use the speech to text feature on a phone. Between auto correct and incorrect words, it\u2019s definitely not perfect. In fact, what is most frustrating is that it doesn\u2019t learn. I always have to correct my daughters\u2019 names as well as that of my engineering VP\u2013EVERY TIME! This is a slightly different use case than query response, but it\u2019s similar. Typically short sentences, real time transcription and the errors are because the context is free form so there isn\u2019t the possibility of comprehensive training.[\/vc_column_text][\/vc_column][vc_column column_padding=”no-extra-padding” column_padding_tablet=”inherit” column_padding_phone=”inherit” column_padding_position=”all” column_element_spacing=”default” background_color_opacity=”1″ background_hover_color_opacity=”1″ column_shadow=”none” column_border_radius=”none” column_link_target=”_self” column_position=”default” gradient_direction=”left_to_right” overlay_strength=”0.3″ width=”1\/3″ tablet_width_inherit=”default” tablet_text_alignment=”default” phone_text_alignment=”default” animation_type=”default” bg_image_animation=”none” border_type=”simple” column_border_width=”none” column_border_style=”solid”][image_with_animation image_url=”16009″ image_size=”full” animation_type=”entrance” animation=”Fade In” hover_animation=”none” alignment=”” border_radius=”none” box_shadow=”none” image_loading=”default” max_width=”125%” max_width_mobile=”default”][\/vc_column][\/vc_row][vc_row type=”full_width_background” full_screen_row_position=”middle” column_margin=”default” column_direction=”default” column_direction_tablet=”default” column_direction_phone=”default” bg_color=”#ffffff” scene_position=”center” top_padding=”5%” constrain_group_1=”yes” bottom_padding=”5%” text_color=”dark” text_align=”left” row_border_radius=”none” row_border_radius_applies=”bg” overflow=”visible” advanced_gradient_angle=”0″ overlay_strength=”0.3″ gradient_direction=”left_to_right” shape_divider_position=”bottom” bg_image_animation=”none” gradient_type=”default” shape_type=””][vc_column column_padding=”no-extra-padding” column_padding_tablet=”inherit” column_padding_phone=”inherit” column_padding_position=”all” column_element_spacing=”default” background_color_opacity=”1″ background_hover_color_opacity=”1″ column_shadow=”none” column_border_radius=”none” column_link_target=”_self” column_position=”default” gradient_direction=”left_to_right” overlay_strength=”0.3″ width=”2\/3″ tablet_width_inherit=”default” tablet_text_alignment=”default” phone_text_alignment=”default” animation_type=”default” bg_image_animation=”none” border_type=”simple” column_border_width=”none” column_border_style=”solid”][vc_column_text]\n
How TranscribeMe Uses ASR<\/h2>\n
The TranscribeMe use case for ASR is neither of these. Eg, \u201cOk Google! Listen to this one hour audio file and transcribe it with timestamps for every speaker change.\u201d As they say colloquially, \u201cthat dog don\u2019t hunt.\u201d Why not? That\u2019s not the use case for Google.<\/p>\n
Simplistically, the ASR industry breaks down into two use cases, query\/response and audio to text transcription. TranscribeMe continually tests vendors’ speech engines and the big 3 or 4 are never at the top of the list in terms of word error rate for our use case\u2013and that makes sense\u2013audio to text, where audio, not \u2018spoken speech\u2019 is not their design target.<\/p>\n
An example of a TranscribeMe virtual request might be, \u201cTranscribe this six hour legal deposition with five speakers using the state of Iowa output format and include speaker IDs and speaker change timestamps.\u201d Well, truth time, no ASR engine is going to get that right. But some may be better than others.<\/p>\n
So that\u2019s where ASR analysis becomes more sophisticated. We\u2019re not simply looking at word error rates but at other factors, such as which engine punctuates or capitalizes best? Which works best w\/ crosstalk? Which is stellar with single speaker or multichannel vs that which can handle multiple speakers on a single channel?<\/p>\n
Why do these qualifications matter? Because the speech engine is not going to produce the final output that will be acceptable to the customer. Maybe it will produce output that\u2019s 90% correct\u2013that\u2019s pretty good. What if your car worked 90% of the time\u2013pretty good or totally unacceptable?[\/vc_column_text][\/vc_column][vc_column column_padding=”no-extra-padding” column_padding_tablet=”inherit” column_padding_phone=”inherit” column_padding_position=”all” column_element_spacing=”default” background_color_opacity=”1″ background_hover_color_opacity=”1″ column_shadow=”none” column_border_radius=”none” column_link_target=”_self” column_position=”default” advanced_gradient_angle=”0″ gradient_direction=”left_to_right” overlay_strength=”0.3″ width=”1\/3″ tablet_width_inherit=”default” tablet_text_alignment=”default” phone_text_alignment=”default” animation_type=”default” bg_image_animation=”none” border_type=”simple” column_border_width=”none” column_border_style=”solid” gradient_type=”default”][image_with_animation image_url=”604″ image_size=”full” animation_type=”entrance” animation=”Fade In” hover_animation=”none” alignment=”” border_radius=”none” box_shadow=”none” image_loading=”default” max_width=”125%” max_width_mobile=”default”][\/vc_column][\/vc_row][vc_row type=”full_width_background” full_screen_row_position=”middle” column_margin=”default” column_direction=”default” column_direction_tablet=”default” column_direction_phone=”default” bg_color=”#f7f7f7″ scene_position=”center” top_padding=”5%” constrain_group_1=”yes” bottom_padding=”5%” text_color=”dark” text_align=”left” row_border_radius=”none” row_border_radius_applies=”bg” overflow=”visible” advanced_gradient_angle=”0″ overlay_strength=”0.3″ gradient_direction=”left_to_right” shape_divider_position=”bottom” bg_image_animation=”none” gradient_type=”default” shape_type=””][vc_column column_padding=”no-extra-padding” column_padding_tablet=”inherit” column_padding_phone=”inherit” column_padding_position=”all” column_element_spacing=”default” background_color_opacity=”1″ background_hover_color_opacity=”1″ column_shadow=”none” column_border_radius=”none” column_link_target=”_self” column_position=”default” gradient_direction=”left_to_right” overlay_strength=”0.3″ width=”2\/3″ tablet_width_inherit=”default” tablet_text_alignment=”default” phone_text_alignment=”default” animation_type=”default” bg_image_animation=”none” border_type=”simple” column_border_width=”none” column_border_style=”solid”][vc_column_text]\n
No ASR Engine\u2019s Output is Perfect<\/h2>\n
No ASR engine with a few caveats can produce output that will be acceptable to the customer as a finished product. The ASR engine produces an output that then requires human review and correction for completion. And that human in the loop dictates which engine we use for various customer and use cases\u2013those distinctions I mentioned above: dial up the ASR that excels at single speaker clear audio; or dial up the ASR that accurately timestamps speaker changes; or we need the engine that doesn\u2019t insert gibberish when it doesn\u2019t understand the audio.<\/p>\n
In summary, the TranscribeMe use case requires different engines for different types\/qualities of audio and for specific use cases. Since we don\u2019t build our own ASR we can shop and use any vendor that fits our needs and provides the best output for human review and correction.<\/p>\n
I mentioned a caveat where there are cases that a one pass ASR output can satisfy customer requirements and in our case we have a customer who does further analytics on the ASR output. That analysis may be keyword spotting or sentiment analysis, or other.<\/p>\n
As an aside, be wary of any company using their own home grown ASR to process files\u2013one size does not fit all and companies that do produce their own ASRs continually narrow the niches where they play.<\/p>\n
Do you have some examples of ways where you have found ASR technology challenging with a project you have worked on? We\u2019d love to know. Are you looking for a company like TranscribeMe to help you with any of your Transcription<\/a> or AI Datasets and Machine Learning<\/a> needs?<\/p>\n