StarZero All shows Try the podcast clipper →

Sep 14, 2015 · 21m · mad

Evan Macmillan, Gridspace // How Machines Enter The Conversation (Hosted by FirstMark Capital)

Evan Macmillan · 15m spoken Matt Turck · 1m spoken George Davies · 19s spoken
0:00 / 0:00
▶ Watch on YouTube →

gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions

In this DataDrivenNYC presentation, Gridspace Founder and CEO Evan MacMillan discusses the technical evolution of speech recognition, the distinct challenges of processing conversational voice data, and how task-driven enterprise software transforms unstructured spoken communication into actionable business intelligence.

How this conversation actually went

Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Matt holds 7.1% of the talking time here. How this is scored →

Matt as informed peer 1.0 Guest teaching 3.8 Guest disagreement 0.5 Matt pushing back 0.5
05100:0010:0020:001:17–6:17 · Matt as informed peer 0/10 Data-Driven Analysis of Past DataDrivenNYC Speakers This segment is a presentation monologue by Evan Macmillan analyzing transcript data from past DataDrivenNYC talks, meaning the host does not participate and host metrics are 0. Evan presents light educational insights on how VCs like Matt Turck talk about business versus technical topics.6:17–8:20 · Matt as informed peer 0/10 Advances in Speech Recognition and Task-Driven Interfaces Evan continues his presentation uninterrupted, explaining the technical complexities of audio pipelines compared to computer vision pipelines. Because Matt Turck is not speaking, host expertise and pushback remain at 0.8:20–11:06 · Matt as informed peer 0/10 Product Demonstrations of Gridspace Memo and SIFT Evan demonstrates Gridspace Memo and SIFT while giving a historical reference to a 1952 Bell Labs speech recognition experiment. The host remains silent throughout the product walkthrough.11:06–21:01 · Matt as informed peer 4/10 Moderated Q&A on Enterprise Voice Systems and Machine Learning Matt Turck hosts a Q&A session, posing knowledgeable questions about data engineering pipelines and telephony integration. Evan and various audience members engage in a constructive, collaborative discussion with minor playful teasing regarding slide data.1:17–6:17 · Guest teaching 3/10 Data-Driven Analysis of Past DataDrivenNYC Speakers This segment is a presentation monologue by Evan Macmillan analyzing transcript data from past DataDrivenNYC talks, meaning the host does not participate and host metrics are 0. Evan presents light educational insights on how VCs like Matt Turck talk about business versus technical topics.6:17–8:20 · Guest teaching 4/10 Advances in Speech Recognition and Task-Driven Interfaces Evan continues his presentation uninterrupted, explaining the technical complexities of audio pipelines compared to computer vision pipelines. Because Matt Turck is not speaking, host expertise and pushback remain at 0.8:20–11:06 · Guest teaching 4/10 Product Demonstrations of Gridspace Memo and SIFT Evan demonstrates Gridspace Memo and SIFT while giving a historical reference to a 1952 Bell Labs speech recognition experiment. The host remains silent throughout the product walkthrough.11:06–21:01 · Guest teaching 4/10 Moderated Q&A on Enterprise Voice Systems and Machine Learning Matt Turck hosts a Q&A session, posing knowledgeable questions about data engineering pipelines and telephony integration. Evan and various audience members engage in a constructive, collaborative discussion with minor playful teasing regarding slide data.1:17–6:17 · Guest disagreement 0/10 Data-Driven Analysis of Past DataDrivenNYC Speakers This segment is a presentation monologue by Evan Macmillan analyzing transcript data from past DataDrivenNYC talks, meaning the host does not participate and host metrics are 0. Evan presents light educational insights on how VCs like Matt Turck talk about business versus technical topics.6:17–8:20 · Guest disagreement 0/10 Advances in Speech Recognition and Task-Driven Interfaces Evan continues his presentation uninterrupted, explaining the technical complexities of audio pipelines compared to computer vision pipelines. Because Matt Turck is not speaking, host expertise and pushback remain at 0.8:20–11:06 · Guest disagreement 0/10 Product Demonstrations of Gridspace Memo and SIFT Evan demonstrates Gridspace Memo and SIFT while giving a historical reference to a 1952 Bell Labs speech recognition experiment. The host remains silent throughout the product walkthrough.11:06–21:01 · Guest disagreement 2/10 Moderated Q&A on Enterprise Voice Systems and Machine Learning Matt Turck hosts a Q&A session, posing knowledgeable questions about data engineering pipelines and telephony integration. Evan and various audience members engage in a constructive, collaborative discussion with minor playful teasing regarding slide data.1:17–6:17 · Matt pushing back 0/10 Data-Driven Analysis of Past DataDrivenNYC Speakers This segment is a presentation monologue by Evan Macmillan analyzing transcript data from past DataDrivenNYC talks, meaning the host does not participate and host metrics are 0. Evan presents light educational insights on how VCs like Matt Turck talk about business versus technical topics.6:17–8:20 · Matt pushing back 0/10 Advances in Speech Recognition and Task-Driven Interfaces Evan continues his presentation uninterrupted, explaining the technical complexities of audio pipelines compared to computer vision pipelines. Because Matt Turck is not speaking, host expertise and pushback remain at 0.8:20–11:06 · Matt pushing back 0/10 Product Demonstrations of Gridspace Memo and SIFT Evan demonstrates Gridspace Memo and SIFT while giving a historical reference to a 1952 Bell Labs speech recognition experiment. The host remains silent throughout the product walkthrough.11:06–21:01 · Matt pushing back 2/10 Moderated Q&A on Enterprise Voice Systems and Machine Learning Matt Turck hosts a Q&A session, posing knowledgeable questions about data engineering pipelines and telephony integration. Evan and various audience members engage in a constructive, collaborative discussion with minor playful teasing regarding slide data.

speaking balance: gold is Matt, purple is the guest (3 minute bins)

0:00 · Matt 0% · guest 100%0:00 · Matt 0% · guest 100%3:00 · Matt 0% · guest 100%3:00 · Matt 0% · guest 100%6:00 · Matt 0% · guest 100%6:00 · Matt 0% · guest 100%9:00 · Matt 24.2% · guest 75.8%9:00 · Matt 24.2% · guest 75.8%12:00 · Matt 11.1% · guest 88.9%12:00 · Matt 11.1% · guest 88.9%15:00 · Matt 7.5% · guest 92.5%15:00 · Matt 7.5% · guest 92.5%18:00 · Matt 7.4% · guest 92.6%18:00 · Matt 7.4% · guest 92.6%21:00 · Matt 0% · guest 0%21:00 · Matt 0% · guest 0%
Sharpest disagreement ▶ 20:34 Audience member calls out unlabelled data cluster

An audience questioner directly challenges Evan on an unlabelled cluster in his analysis slide, prompting Evan to admit he was caught.

 Hardest push from Matt ▶ 13:01 Host presses beyond telephony incumbents

Matt Turck pushes Evan to clarify if Gridspace is looking beyond standard telecommunication providers like Cisco and Avaya into media content.

 Biggest teaching moment ▶ 16:06 Explaining limitations of voice sentiment analysis

Evan educates the audience on why signal-level sentiment analysis is impractical due to the extreme difficulty of establishing ground truth for human emotions.

 Matt holds his own ▶ 11:35 Host demonstrates technical understanding of data ingestion

Matt Turck highlights his domain knowledge by asking a targeted question about the infrastructure and data engineering required to ingest voice audio in an enterprise setting.
the scores for every segment, with the reasoning behind each
ChapterTopicMatt as informed peerGuest teachingGuest disagreementMatt pushing backWhy
Data-Driven Analysis of Past DataDrivenNYC Speakers 0300 This segment is a presentation monologue by Evan Macmillan analyzing transcript data from past DataDrivenNYC talks, meaning the host does not participate and host metrics are 0. Evan presents light educational insights on how VCs like Matt Turck talk about business versus technical topics.
Advances in Speech Recognition and Task-Driven Interfaces 0400 Evan continues his presentation uninterrupted, explaining the technical complexities of audio pipelines compared to computer vision pipelines. Because Matt Turck is not speaking, host expertise and pushback remain at 0.
Product Demonstrations of Gridspace Memo and SIFT 0400 Evan demonstrates Gridspace Memo and SIFT while giving a historical reference to a 1952 Bell Labs speech recognition experiment. The host remains silent throughout the product walkthrough.
Moderated Q&A on Enterprise Voice Systems and Machine Learning 4422 Matt Turck hosts a Q&A session, posing knowledgeable questions about data engineering pipelines and telephony integration. Evan and various audience members engage in a constructive, collaborative discussion with minor playful teasing regarding slide data.

Statements from this episode (6)

Assertion Not checkable as stated
People speak 100 times more words than they write in email
“So people speak about, ah, a hundred times more words, ah, than they write in email.”
Evan Macmillan Sep 14, 2015 ▶ 3:35
Insight
Macmillan: The best interfaces for voice output are task-driven
“I think the best interfaces are task-driven.”
Evan Macmillan Sep 14, 2015 ▶ 8:18
Assertion Supported
A 1952 Bell Labs voice system reached 97-99% accuracy
“Actually the, ah, the recognition rate of this machine was 97 to 99%.”
Evan Macmillan Sep 14, 2015 ▶ 10:08
Assertion Supported
Cisco and Avaya control the majority of enterprise voice data infrastructure
“So when we're talking about voice data, there's a few companies that control the pipes and those companies are Cisco and Viya, and they have a pretty large footprint in contact center, and then also in unified conferencing.”
Evan Macmillan Sep 14, 2015 ▶ 11:55
Assertion Not checkable as stated
Most YouTube advertisers do not know what video content they target
“Most YouTube, I mean, advertisers don't really know what's going on in the YouTube videos they advertise against.”
Evan Macmillan Sep 14, 2015 ▶ 13:27
Assertion Not checkable as stated
English speech recognition is harder than Romance languages due to orthography
“English was actually harder than many romantic languages because the orthography in English, it means it's like how, how words get written and then said are, there's a better connection in romantic languages than English.”
Evan Macmillan Sep 14, 2015 ▶ 17:35
Made with StarZero

Turn any episode into a week of clips.

This entire site, over 400 conversations transcribed, diarized, checked and made playable, runs on the StarZero media pipeline. Drop in your own episode and the podcast clipper finds the moments worth sharing, cuts them, captions them, and reframes them for every feed.

Try the podcast clipper Learn more