This tech could have revolutionized the voice-over industry (feat. Al Lowe)
Hi, I'm the Space Quest Historian. Back when I used to do the Space Quest Historian podcast I did an episode where I interviewed Al Lowe, creator of Donald Duck's Playground, and also some other game whose name escapes me at the moment. The thing about this interview is you've never heard it. No one has, not even me. And that's because I did this interview way back at a time when recording computer audio was a finicky and cumbersome process, at least to me, which led me to put my trust in a third party Skype recorder application. Big fucking mistake on my part.
What ended up happening was the recorder software recorded about 30 seconds of the call, then had a stroke and completely forgot what it was put on this earth to do. And unfortunately it didn't bother telling anyone. So I carried on having the most delightful and insightful conversation with one of adventure games’ royalty for over an hour, blissfully unaware that the Skype recorder had checked out moments after it started and ended up with a 30 second recording that consisted entirely of me and Al saying hello to each other.
And it's an absolute fucking tragesty that that happened, because Al dropped some trivia on me that, to my knowledge, no one else has had the privilege of knowing. And even if this has since been recounted by the man himself in subsequent interviews, I'm pretty sure I was the one who got to hear it first. So yay me and fuck me.
But I'm happy to say that finally this injustice of catastrophic computerized calamity has been rectified. Fast forward to today in this year of our foul corporate overlords of 2024, I once again had the pleasure and privilege of chatting to Mr. Al... (tee-hee)... Al Lowe. That's my Larry impression, in case that wasn’t obvious. ...on a completely different topic for a completely different video I'm working on, but I couldn't pass up the opportunity to ask Mr. Lowe to recount
the juicy knowledge he imparted on me those many, many years ago. And luckily he obliged. So now, nearly a decade late... ...but, as Larry would have said, better late than never... I can finally tell you about this one piece of amazing tech that Sierra On-Line had back in the day. One that was sadly lost to time because of corporate greed and incompetence and one that sadly only lives on today in the scraps of documentation for the tech, which I am fortunately in possession of and will show you later in this video, and in the memories of people who were there to see it back in the day, like Mr. Al Lowe. Now, as the title of this video suggests, this particular tech has to do with voiceovers.
While Sierra were at the forefront of CD-ROM technology in the early nineties with King’s Quest V premiering on the shiny disc-shaped medium in 1992, a full year before Myst and The 7th Guest properly legitimized the medium, their earliest forays into CD-ROM involved cutting a lot of corners, predominantly achieved by getting people who already worked at the offices to do voiceovers for the games. Case in point, the aforementioned King’s Quest V features the voice talents of Quest for Glory designer Lori Cole, King's Quest designer Roberta Williams, and most famously programmer Richard Aaronson as Cedric the Owl. And they were at the time easily upstaged by their contemporaries, most notably their primary rival, LucasArts, who actually went to the trouble of hiring real professional voice actors for their early CD-ROM titles like Day of the Tentacle. But Sierra did eventually realize that reading stuff aloud is something best left to people who do that sort of shit for a living.
And Al Lowe’s games were at the forefront of that. I mean, say what you will, if you indeed have anything to say Al Lowe’s games, one inarguable thing is that every CD-ROM version of his games all had fantastic voice acting. I mean, Freddy Pharkas had Cam Clarke as the titular character.
Leisure Suit Larry had the inimitable Jan Rabson, may his sleazy soul rest in peace, and Torin’s Passage had... well, actually, I don't recognize any of these names, but they still fucking delivered. So they were clearly professionals. But recording voiceovers for CD-ROM games is a lot of work, and I mean a lot.
And that's why, if you play Freddy Pharkas on CD-ROM, you're actually getting a super truncated version of the script where a lot of the dialog got excised because the script was just too big. Thanks, Josh Mandel. Including, unfortunately, a lot of the hints to the puzzles which inadvertently made the game much more difficult in the CD-ROM version. But that's not the point. The point is: Recording voiceovers doesn't just involve getting some yahoo in front of the microphone and have them read out a script, like I'm doing right now, although that's usually a good first step.
But once you're done with that, you're faced with the arduous and mind-meltingly tedious task of having to cut up all the individual lines into separate audio files and meticulously giving them the proper file names so the game will know which audio file to play at any given time. And any hobbyist who has ever made a game in Adventure Game Studio and thought, “Hey, why don't I include voice acting as well?” You'll know exactly the kind of 7th circle of Hell tedium I’m talking about here because this process takes for-fucking-ever. But...
Sierra came up with a clever solution to this problem with some fancy in-studio magic. And from here, I'll just let Mr. Al Lowe himself explain the sorcery they came up with and how it all worked. (AL LOWE) When we started doing CD-ROM games and we realized we could have voiceover, one of the engineers, one of the systems engineers, said, “Gosh, breaking out lines with an editor and a mouse and clipping this and doing this thing and then saving that as some magic number someplace and then telling...
rewriting the code so it would know to look for that magic number. Um, that's horrible. We can't do that.” So instead, he said, DOS has eight letters and three extensions. Right? An extension with three bytes.
So he came up... He took the first two... I can't bend those fingers. He put the first two... Yeah, that's good. ...and assigned those to a 36 character alphabet. So he took all the letters and numbers and so he made everything base 36.
By using two bytes, two characters, that gives you 36 times 36 possible combinations... ...or maybe one less than that, probably. But anyway, what is that number? Do you know? It’s big. (SQH) I am not mathematically inclined, but it sounds like a lot.
(AL) It’s a couple thousand. It's... it's up there. Somebody’ll answer that. Put it in the comments below. So using base 36, he assigned every room two characters, so we could have more than 36 rooms, of course. And then within that room there were conditions.
And so the conditions were two bytes and the characters were two bytes. And the message for that character was... Ah, shit, I’m missing something. Anyway, it was four groups of two bytes each. So there are billions of combinations. And then the extension was the number of the action... [TRAILS OFF]
And so then once he had that 11 character magic number then whatever you did in a room; in a scene was a unique action. So if you said “look tree,” you know, that would be, well, where are you? Who are you? What are you doing? Who's going to speak? What's the answer? And here’s the message number. That worked great. So he took the code that we wrote for the game and broke it apart so that it would record in another file every possible combination that could call up a message. And then it sorted all those by speaker.
So everything that Larry said, everything that Torin said, was put in a script, one line after another, and it had this magic number off to the side. But nobody ever looked at it. Nobody used that. It was just for the computer. And then we had the text for it. And we had a line number, I think, like, a decimal number that was, you know, line number 2343. But nobody ever wrote that down.
It was just the way... that's how we talked to the talent in the booth. Okay, so... Now he would take this, like, Larry 7 script...
The night before I went to the studios to record all those voiceovers, I ran his program and it produced this gigantic file that I then printed. And it was... I had to buy a special fat binder because it was this much... even thin, cheap paper that Sierra used... ...it was this much paper. And it had every line on it. Now, he also said... I said something, “Well, if we're doing dialogs, we want to know what the other person said.”
(SQH) So there's interrelation... (AL) Oh, yeah. So... So then it was a section of lines for Larry to say that were singletons. By themselves. And then it had all the Larry plus... Marjorie... you know, dialog.
So any time the two of them interacted, or any three people interacted, that was another chapter, and then so forth. And it went through every possible permutation and combination, but only once. So you didn't record the lines twice. And if you did make a mistake when you recorded the somebody and somebody else with Larry, well, their lines would be the same number as that line that you had first. So in case you did record it over, it didn't matter because it would only save as the one magic number. Then we said, “Well, this is going to be a horrible job to break 10,000 or 8000 lines out of an audio file.
How can we handle that?” He said, “Well, there's this new thing called DAT.” “Have you heard of DAT?” (SQH) Digital Audio Tape, for those playing at home. (AL) Yeah.
So he picked one DAT machine... I believe it was a Panasonic... ...that had a serial port. And he studied that and he said, “You know, laptops have a serial port.” Well, they did back then. [LAUGHS] “So if I connected those two serial ports together, could I make the laptop control the DAT machine?” And he did. And then he said, “Well, if the laptop has the script, Al's original script that was this thick, in it as one big text file, couldn't we just jump to page 600 and start recording there and have the DAT machine send the, uh, NAB code, which is... National Association of Broadcast. A timecode, in other words. When you start the tape
it says zero, and then from then on it counts 30ths of a second, I think? Whatever. It counts some magic, you know, type number. And so then he has the program on the laptop when you press the... I think it was the spacebar it stores the magic 11 digit number on a text file and it stores the DAT code of the start point of that. And then when you're done, you touch the spacebar again, when the guy's done reading, and that writes a little thing to the end code, and... I think that's all it records. It’s, like, 11 bytes and then 20 and 20... But it's a tiny little text file.
And so we would put it an hour DAT tape and just record in the studio. And the guy running the laptop would... I was the director and we had the audio engineer do the laptop. So he would go down the list of lines that had to be read.
He would say, “All right, we're ready to read nine line number 1027.” And the guy would flip in his paper, you know, and he’d go, “Oh, okay, I got it. And what am I doing here? All right, I got it.” And then it was ready, and he would touch the spacebar and the guy would read the line, the actor would read the line. And then when he was done, they would touch the spacebar again.
And that was the out point. And then that line was written into the database on a file somewhere. At the end of the day, we would have maybe six DAT tapes full of data like this, and all of it fit in one little data file. But we made six. We made one data file for each tape so not to be confused. And so... at the end of the day, we would copy that data to a floppy, take the six DAT tapes, put them in a FedEx box, and send them back to Oakhurst...
or back to Bellevue on later games. By the time the guy showed up the next morning, they would break out all the lines from the day before, the recording, and the game would just start speaking, as if magic. He also wrote... [LAUGHS] another program. He said, “Well, if I know the start point and the end point... Yeah, but sometimes you'd press the spacebar early or sometimes you'd press it late. And what are you going to do about that?” Well, he wrote a program called Trim that would automatically go through and... [SQH LAUGHS]
[AL RECIPROCATES LAUGH] take out all the silence at the beginning and end of the lines. And then [LAUGHS] we had the problem with... Well, all the characters have a kind of different volume. You know, there was no compression or anything. And so he wrote a program that would go through the entire audio directory. It was called Normalize. (SQH) Mm-hmm.
(AL) You would type “normalize.” And it would go through and bring up the volume of anything too loud, take down the volume, anything to... vice versa. Yeah, I mean, absolutely astonishing.
And what did Sierra do with that? Every time we went into a studio, the audio engineer who worked in the studio in Hollywood would say, “What in the hell is this? What is that? I want that software. Where can I buy that?” What did they do with it? [SQH MAKES WHOOSHING SOUND] (AL) Into the ether. (SQH) Into the ether. Well, I guess we can technically blame CUC for that, but... (AL) Yeah, yeah, yeah.
(SQH) They just shuttled the whole thing. (AL) They had no idea what they had. Now, CUC here refers to the company that bought out Sierra, then became the Cendant Corporation, and eventually blah blah blah blah, massive financial scandal, blah blah, death of the company, blah. I think that story's been told a few times now. Anyway, nowadays we do have similar tools for audio processing.
In fact, there's an online tool I’ll link to in the video description that can automatically cut up an audio file into individual takes and then trim the silences out, which is absolutely free to use. And it works surprisingly well. But for Sierra to have this technology with that system back in the mid-90s was just an absolute gem. And it's not hyperbolic to suggest that it could have been nothing short of revolutionary for the voice-over industry of the time, but it was only used in-house at Sierra.
It was never packaged and marketed for outside use. It was used exclusively by Sierra for their CD-ROM games. And when Sierra went belly-up after the aforementioned financial hullabaloo, all the company's assets, including all of the proprietary in-house tech, along with their irreplaceable artifacts such as original artworks and everything else they used to make their games, were unceremoniously consigned to a dumpster somewhere, never to be seen again.
Which is kind of a shame, I think you'll agree. But anyway, that's the story of the magic tech behind Sierra's CD-ROM games. Just imagine being able to wake up the next day after a recording session and booting up the game you're working on, only to hear the voice actors you just recorded yesterday speaking loud and clear in the game you're making. There's no fiddling with audio levels, no fucking around with file name conventions, no tedious and meticulous, cutting up of different takes. You just boot it up and watch it explode. I’m the Space Quest Historian. There’s more juicy goodness
from my interview with Al Lowe coming in the future. And my voice just broke, but I don’t care. I'm not retaking this line. And soon you'll be able to see the full, uncut interview... tee-hee... on my second channel, the SQH Dumping Grounds.
Please give that a sub, by the way, if you haven't already. [GIGGLES] Enough giggling from me. Until next time, I'll see you around the chronostream. Bye!
2024-05-26 06:32
