Academic Lecture Transcripts
This is a volunteer experimental project transcribing academic lectures and putting the transcripts within a MediaWiki website. The content is from HIST 186 International and Global History since 1945 taught by Daniel Sargent at UC Berkeley in the spring of 2012.
The transcription was done using Plover which is part of the Open Steno Project. I also tried out adding headings, links, notes, references, word definitions, and occasionally embedded images and video.
The wiki is not currently editable by readers like is the case on Wikipedia for most articles, but people can still send me email with corrections or comments. In the subject of the email include at the beginning "AcLeTr:". The username for my Gmail address is david.kit.friedman .
Typos and minor errors can most often be corrected quickly, or if there's a broken link and the new URL is easily obtained that could be a good edit to make to a Wikipedia article or a wiki page. More substantial changes and fixes to this wiki may not be worth it though. Depending on how things go I might not get to it for a few weeks or a month.
I tried to contact Daniel Sargent and other people at UC Berkeley on this transcription work at the end of 2018 and at the beginning of 2019, but I didn't get any response.
There are no plans currently to transcribe any additional classes, but that could nevertheless be a possibility for the future.
Internet Archive Page for HIST 186
(see also Technical Comments)
DavidKitFriedman (talk) 22:19, 24 May 2019 (UTC)
Intro Revised: DavidKitFriedman (talk) 07:03, 15 December 2021 (UTC)
Brief Postscript: My username on Wikipedia is Jjjjjjjjjj (contribs) and in the course of listening to the lectures and doing the transcriptions I did various Wikipedia editing.
DavidKitFriedman (talk) 07:05, 26 May 2019 (UTC)
I just submitted a review of the lecture series which is available on the details page on Internet Archive for the course.
DavidKitFriedman (talk) 20:54, 14 June 2019 (UTC)
For security I changed all the passwords for the accounts on this wiki, but if any of the people to whom I sent login credentials would like to have access or to talk about any changes then feel free to email me at the address mentioned above and include "AcLeTr:" in the subject.
DavidKitFriedman (talk) 06:47, 15 December 2021 (UTC)
2022-04-05: Report on Trying Out Otter.ai automated transcription service
In the course of preparing for a consulting meeting with Wiki Education↗. I recently did a Google search for "audio transcription", and received an ad for Otter.ai↗ automated transcription service...
With a free basic plan and a $12.99 per month pro plan I decided to try Otter.ai using the free basic plan.
The free basic plan offers 600 minutes per month of automated audio transcription done while logged onto the website and with the pro plan one gets 6000 minutes per month.
Unlimited import of audio files is available with the pro plan, but with the free plan one gets a trial of three audio imports with a limit of 30 minutes for each file. Note that these three audio imports is per account and not per month.
I tried it out on Lecture 16 - The Cold War Resurges and the resulting transcript for the first 30 minutes can be found here as a plain text file, and also as a wiki page.
I compared my transcript with the Otter.ai transcript for the first 10 minutes of the recording. In the course of this comparison I found some errors in my transcript which Otter.ai got correct, or other errors, and one can see these instances in the list of contributions starting at 15 March 2022 where I started the edit summary with, "Found while comparing this transcript to an automated one produced by Otter.ai."
The Otter.ai transcription has most of the words correct within lengthy paragraphs, but there can be occasional errors. For example, at the beginning it got "I hear..." instead of "I fear...".
Otter.ai was successfully able to get a variety of proper nouns including Nikita Khrushchev↗, Brezhnev↗, ZiL limousine↗, Bolsheviks↗, Kosygin↗. In the case of Kosygin, for the first name, the spelling "Alexei Kosygin" looks to be more standard in results from ProQuest than "Alexey Kosygin" which is what Otter.ai got though both are listed as romanizations of that particular Russian and Bulgarian first name in the Wikipedia article with the title Alexey.
On the other hand it was not able to get Samizdat↗, Aleksandr Solzhenitsyn↗, or Ostpolitik↗. For Jean-Paul Sartre↗ it got John Paul Sartre. Sometimes it got détente and sometimes it didn't, and similarly sometimes it got Lévy in the case of the name Bernard-Henri Lévy↗ and sometimes it didn't.
Sometimes the sentence construction could be awkward such as with:
- Okay, first, the Soviet Union. What do we need to know about the travails of Soviet style socialism during the 1970s, the Brezhnev years in the history of the Soviet Union, a period that begins towards the late 1960s. It's difficult to put a specific point on the origins of the Brezhnev era, because Khrushchev fall he was ousted in 1964. It's followed by a period of collective leadership.
Which could be changed to:
- Okay, first the Soviet Union. What do we need to know about the travails of Soviet-style socialism during the 1970s?
- The Brezhnev years, in the history of the Soviet Union, a period that begins, towards the late 1960s. It's difficult to put a specific point on the origins of the Brezhnev Era because Khrushchev's fall, he was ousted in 1964, is followed by a period of collective leadership.
Otter.ai offers the capability to edit the produced transcript via the website, and one can play, pause, rewind five seconds, slow down, speed up, using hotkeys as mentioned in this article. Also mentioned there is how the machine learning can learn over time to produce better transcripts. Also available is Otter.ai for Education which I haven't investigated.
When I was doing the transcripts available on this website, as I mentioned at the top of this page, I used stenography, and I also used a website called oTranscribe which similar to Otter.ai has hotkeys for play, rewind, slow down, etc. Using stenography along with oTranscribe I was able to set it up so that when a proper noun was heard that I realized wasn't in the steno dictionary (or perhaps which I didn't remember how to do, or perhaps which I couldn't do quickly enough) I could instead just press a key combination (called a chord in stenography) that inserted "[PNOUN <timestamp>]" where <timestamp> was the timestamp for that proper noun (e.g. (05:25) for 5 minutes and 25 second into the audio file).
It was then possible to do another pass through the transcript, and fill in the proper nouns. Using oTranscribe it was possible to hear the audio at that timestamp by just clicking on the produced link.
I could then also optionally easily insert a link to a Wikipedia article for that proper noun by using a template that I made called {{WPExtract}}.
One could then add that word to the steno dictionary so that the next time it is spoken it could be done without needing to use PNOUN.
I haven't tried out Otter.ai more fully in terms of say for example doing time, accuracy, quality measurements, etc. between a variety of different techniques, but that's something that one could think about.
DavidKitFriedman (talk) 23:28, 5 April 2022 (UTC)
