I think I have figured out one of the reasons why dictation is so difficult for me.
I've known a lot of writers, and some writers will just vomit onto the page. The words come out as they think them, and they don't care if the words aren’t exactly the way they want because they'll just fix it and edit it later.
There is also the type of writer that sits and thinks for a long time until they get the sentence exactly how they want it, and then they type it out, and the sentence is exactly what they wanted without needing any edits.
But I am not like either of those. I type something onto the screen, but I edit it. And I keep fiddling with it until the sentence is how I want it. And then I move onto the next sentence.
I think that for the first two types of writers, dictation is probably easier. The first type will simply speak the first thing in their heads and the dictation will transcribe their words. For the second type, they think carefully about what sentence they want to write, and then speak it, and their dictation ends up being very clean and needs hardly any editing, just like their regular writing on the computer.
But for me, I've been finding that I dictate just like how I write on the computer. I will dictate a sentence, but it won't sound right so I'll dictate the sentence again, and I might dictate two or three more times until it's the way I want.
It ends up needing a lot of cleanup editing later because I have to delete the versions of the sentence that I don't want. It also means that a large chunk of my dictation is deleted, so it's misleading how many words per hour I write when I dictate.
I've started learning how to think of a sentence in my head before I speak it, but sometimes that's difficult and I default to just spitting out the sentence and then repeating it until it's the way I want it.
I think for writers like me, this is why dictation is so hard. I know how to speak, and I know how to write, but dictating is something else entirely because I am forced to think about the sentence more before speaking it, and that is not a method of writing that I'm used to.
Recently, AI tools have solved some of my problems for me.
Before, I would just dictate the way I write, and continue to dictate a phrase or sentence, then repeat it until it's the way I want it. But that ended up taking a lot of time for cleanup editing, which could be very frustrating, especially because I really dislike self-editing.
I thought that the only option was to to learn how to think through a sentence in my mind before I dictate it. That may have been one main reason why dictation has taken me so long to get used to. The only other option was to simply dictate the way I write and resign myself to the fact that I’d have a lot of cleanup later.
When I tried AI transcription, one of the reasons I liked it was because sometimes it would clean up the transcription for me, but one thing I disliked about AI transcription is that sometimes it would delete notes that I leave for myself.
However, recently, Thomas Umstattd created an AI tool called Chapterizer (one of the tools on his Patron Toolbox, which requires you to subscribe to his Patreon) that cleans up dictation WITHOUT deleting the notes I made to myself, and also without changing my writing as it cleans it up! I’ve compared before and after for all the pieces I’ve thrown through the Chapterizer so far, and it’s FANTASTIC. I think my cleanup woes are solved.
Cleanup still takes some time because the Chapterizer doesn’t catch all my repetition. But Chapterizer has cut down my cleanup time by at least half, which I think is significant.
I’ve also realized that the majority of time I take during cleanup is in rearranging sentences if I was dictating and then remembered something else I should have added earlier. Since I’m dictating into my digital voice recorder, I have to just leave a note to myself and keep going. Following the instructions of my notes during cleanup takes the most time, I think.
It might be easier if I were dictating directly into Scrivner or Pages on my phone. I have been using my digital voice recorder because I write a lot faster when I use it, specifically because I can’t see what I’ve written and I have to just keep going. If I can see my sentences, my dictation words per hour rate goes way down.
I also have my notes on what happens in the scene in my phone, and switching between apps on my phone was a pain. But I just realized that I can copy my notes into Dragon Anywhere and dictate directly above my notes, so I can refer to the notes as I dictate the prose without needing to leave the app. I might try that, especially now that I can use Chapterizer to clean up my repetition, rather than fixing it as I dictate into Dragon Anywhere.
My only concern then will be the quality of the transcription. I know I get pretty accurate transcription using my digital voice recorder and Dragon on my computer, while using my phone microphone is not as good. But maybe it’s worth it to cut down on the editing I’ll have to do later if I keep writing sentences out of order.
I’m still not sure if dictating the way I write (by repeating the sentence over and over until it's the way I want it) will help me to dictate faster. At the moment, I’m only writing a little more than 1000 words an hour if you count the cleanup time (even with cleanup time being reduced by using Chapterizer).
I suppose I’ll just have to keep working on what method of dictation works best—digital voice recorder, or live transcription in the Dragon Anywhere app. And maybe I’ll do a combination of the two, depending on where and when I’m dictating.
It’s kind of interesting that even though I’ve been writing for so many years, there’s still so much I can learn and improve. I’d love to get better and faster at dictation so that I can get more words written each day.
***
After I wrote the above, I figured out another way to help me when I dictate into my digital voice recorder. It still doesn’t help if I write sentences out of order, but it does help when I misspeak.
I read a book on dictation by Michael La Ronn, and he has an interesting tip for when you make a mistake while dictating. He says “Pikachu period” when he’s made a mistake. Later, you can do a search for Pikachu and delete any sentence with that word in it. What he did was create a Microsoft Word macro to do it automatically for him.
I don’t use Microsoft Word, but I realized that with ChatGPT, I could ask it to write me a program to do that on my Mac. It was very helpful and after some testing and a few back and forth iterations, I finally got an app made on my computer that will automatically remove Pikachu sentences!
Here’s instructions and the script for the app. Sorry, but this only works for Mac users.
Pikachu Method:
(Method from M.L. Ronn)
When you are dictating into a recording app or digital recorder, if you make a mistake, say
Pikachu period
Then repeat the entire sentence.
After transcription of the MP3 file, save the transcription as a .txt file.
To remove Pikachu sentences, I asked ChatGPT to make an app for me.
🧰 How to Use in Automator
1. Open Automator → New Document → Application
2. Search for Run Shell Script and drag it into the workflow.
3. Set “Pass input” to: as arguments and set “Shell” to: /bin/bash
4. Copy and paste the script below..
5. Save the application (e.g. Remove Pikachu Sentences.app) anywhere you like (Desktop is convenient).
6. Now drag .txt files directly onto your app. It will take a few seconds and then create a “cleaned” txt file and also a file showing what was removed..
Here’s the script to copy and paste:
I've known a lot of writers, and some writers will just vomit onto the page. The words come out as they think them, and they don't care if the words aren’t exactly the way they want because they'll just fix it and edit it later.
There is also the type of writer that sits and thinks for a long time until they get the sentence exactly how they want it, and then they type it out, and the sentence is exactly what they wanted without needing any edits.
But I am not like either of those. I type something onto the screen, but I edit it. And I keep fiddling with it until the sentence is how I want it. And then I move onto the next sentence.
I think that for the first two types of writers, dictation is probably easier. The first type will simply speak the first thing in their heads and the dictation will transcribe their words. For the second type, they think carefully about what sentence they want to write, and then speak it, and their dictation ends up being very clean and needs hardly any editing, just like their regular writing on the computer.
But for me, I've been finding that I dictate just like how I write on the computer. I will dictate a sentence, but it won't sound right so I'll dictate the sentence again, and I might dictate two or three more times until it's the way I want.
It ends up needing a lot of cleanup editing later because I have to delete the versions of the sentence that I don't want. It also means that a large chunk of my dictation is deleted, so it's misleading how many words per hour I write when I dictate.
I've started learning how to think of a sentence in my head before I speak it, but sometimes that's difficult and I default to just spitting out the sentence and then repeating it until it's the way I want it.
I think for writers like me, this is why dictation is so hard. I know how to speak, and I know how to write, but dictating is something else entirely because I am forced to think about the sentence more before speaking it, and that is not a method of writing that I'm used to.
Recently, AI tools have solved some of my problems for me.
Before, I would just dictate the way I write, and continue to dictate a phrase or sentence, then repeat it until it's the way I want it. But that ended up taking a lot of time for cleanup editing, which could be very frustrating, especially because I really dislike self-editing.
I thought that the only option was to to learn how to think through a sentence in my mind before I dictate it. That may have been one main reason why dictation has taken me so long to get used to. The only other option was to simply dictate the way I write and resign myself to the fact that I’d have a lot of cleanup later.
When I tried AI transcription, one of the reasons I liked it was because sometimes it would clean up the transcription for me, but one thing I disliked about AI transcription is that sometimes it would delete notes that I leave for myself.
However, recently, Thomas Umstattd created an AI tool called Chapterizer (one of the tools on his Patron Toolbox, which requires you to subscribe to his Patreon) that cleans up dictation WITHOUT deleting the notes I made to myself, and also without changing my writing as it cleans it up! I’ve compared before and after for all the pieces I’ve thrown through the Chapterizer so far, and it’s FANTASTIC. I think my cleanup woes are solved.
Cleanup still takes some time because the Chapterizer doesn’t catch all my repetition. But Chapterizer has cut down my cleanup time by at least half, which I think is significant.
I’ve also realized that the majority of time I take during cleanup is in rearranging sentences if I was dictating and then remembered something else I should have added earlier. Since I’m dictating into my digital voice recorder, I have to just leave a note to myself and keep going. Following the instructions of my notes during cleanup takes the most time, I think.
It might be easier if I were dictating directly into Scrivner or Pages on my phone. I have been using my digital voice recorder because I write a lot faster when I use it, specifically because I can’t see what I’ve written and I have to just keep going. If I can see my sentences, my dictation words per hour rate goes way down.
I also have my notes on what happens in the scene in my phone, and switching between apps on my phone was a pain. But I just realized that I can copy my notes into Dragon Anywhere and dictate directly above my notes, so I can refer to the notes as I dictate the prose without needing to leave the app. I might try that, especially now that I can use Chapterizer to clean up my repetition, rather than fixing it as I dictate into Dragon Anywhere.
My only concern then will be the quality of the transcription. I know I get pretty accurate transcription using my digital voice recorder and Dragon on my computer, while using my phone microphone is not as good. But maybe it’s worth it to cut down on the editing I’ll have to do later if I keep writing sentences out of order.
I’m still not sure if dictating the way I write (by repeating the sentence over and over until it's the way I want it) will help me to dictate faster. At the moment, I’m only writing a little more than 1000 words an hour if you count the cleanup time (even with cleanup time being reduced by using Chapterizer).
I suppose I’ll just have to keep working on what method of dictation works best—digital voice recorder, or live transcription in the Dragon Anywhere app. And maybe I’ll do a combination of the two, depending on where and when I’m dictating.
It’s kind of interesting that even though I’ve been writing for so many years, there’s still so much I can learn and improve. I’d love to get better and faster at dictation so that I can get more words written each day.
***
After I wrote the above, I figured out another way to help me when I dictate into my digital voice recorder. It still doesn’t help if I write sentences out of order, but it does help when I misspeak.
I read a book on dictation by Michael La Ronn, and he has an interesting tip for when you make a mistake while dictating. He says “Pikachu period” when he’s made a mistake. Later, you can do a search for Pikachu and delete any sentence with that word in it. What he did was create a Microsoft Word macro to do it automatically for him.
I don’t use Microsoft Word, but I realized that with ChatGPT, I could ask it to write me a program to do that on my Mac. It was very helpful and after some testing and a few back and forth iterations, I finally got an app made on my computer that will automatically remove Pikachu sentences!
Here’s instructions and the script for the app. Sorry, but this only works for Mac users.
Pikachu Method:
(Method from M.L. Ronn)
When you are dictating into a recording app or digital recorder, if you make a mistake, say
Pikachu period
Then repeat the entire sentence.
After transcription of the MP3 file, save the transcription as a .txt file.
To remove Pikachu sentences, I asked ChatGPT to make an app for me.
🧰 How to Use in Automator
1. Open Automator → New Document → Application
2. Search for Run Shell Script and drag it into the workflow.
3. Set “Pass input” to: as arguments and set “Shell” to: /bin/bash
4. Copy and paste the script below..
5. Save the application (e.g. Remove Pikachu Sentences.app) anywhere you like (Desktop is convenient).
6. Now drag .txt files directly onto your app. It will take a few seconds and then create a “cleaned” txt file and also a file showing what was removed..
Here’s the script to copy and paste:
for f in "$@"
do
python3 <<EOF
import re
import os
input_file = "$f"
base_name = os.path.splitext(input_file)[0]
output_file = base_name + "_cleaned.txt"
log_file = base_name + "_removed.log"
with open(input_file, "r", encoding="utf-8") as file:
text = file.read()
# Match sentences ending in "Pikachu." (case-insensitive), ending with . and followed by whitespace or end
pattern = r'(?<!\\w)([^.!?]*?\\bPikachu\\.)(?=\\s|$)'
removed_sentences = re.findall(pattern, text, flags=re.IGNORECASE)
with open(log_file, "w", encoding="utf-8") as log:
for sentence in removed_sentences:
log.write(sentence.strip() + "\\n")
cleaned_text = re.sub(pattern, '', text, flags=re.IGNORECASE).strip()
with open(output_file, "w", encoding="utf-8") as out:
out.write(cleaned_text)
print(f"Cleaned version: {output_file}")
print(f"Removed sentences saved to: {log_file}")
EOF
done
Comments
Post a Comment