On January 31, I presented a webinar about Express Scribe, a free transcription program I've been using to transcribe content for clients who need their audio recordings time coded and translated. This webinar session, "Express Scribe: Free Audio Transcription Software," was offered through Proz.com.
We covered the following topics during the live session:
- What is Express Scribe
- Technical Specifications
- System Requirements
- Compatible Hardware
- Formats Supported
- Let's Get to Work!
- Loading Audio Files
- Using hot keys to...
- ...control audio playback
- ...add time stamps
- Transcribing Original Audio
- Time to Translate it!
- Ideas on Project Negotiation
Below you'll find questions that were asked during the webinar, as well as some other comments that were sent to me by email. If you purchase access to the video, have questions after watching the material, and don't see your concerns addressed here, feel free to contact me and contribute to this Q&A list.
How can I start looking for transcription clients?
Just as you would go about finding translation clients, you'd also need to look for people who need transcription services. Think about the areas that would have a good demand. Brainstorm where there is a need for your services. Put your resume together, start practicing, and do the legwork of actually researching your potential client base. You can also pay attention to job boards, like Proz.com or other freelancing websites where people can post information on their project and connect with professionals offering their services.
Three main fields come to mind: Medical, Legal, and Academic. There can be doctors and nurses who recorded information about a patient, instead of putting pen to paper, and they'll need that content transcribed. There are lawyers who have recorded their conversation with their client, or have recordings about an actual court case. There are professors who record their lectures and will need them in text format.
How can I get hands-on practice transcribing audio files?
As mentioned earlier, you can use the audio files available on the Express Scribe website. You can also get creative and record yourself reading a page of a book or a news article, then transcribe what you've recorded. You can even use your TV! Let's say you turn on CNN and use your phone to record Anderson Cooper talking for a few minutes. You can then use the audio file to practice your transcription skills.
Should I correct mistakes made by the speaker on the audio file?
That's a great question! It really depends on the purpose of your project. Let's say you're transcribing something of legal nature, and someone says something wrong, but it's actually a slip of tongue that reveals the truth about a court case! You definitely want to be completely literal in a case like that, so as to not interfere with the legal process. If it's a lecture and the speaker makes a grammar mistake, but will need an accurate sentence in order to publish a scientific article or a book, for example, you could go ahead and correct that.
What happens when the audio quality is bad?
It is definitely a challenge! What I can tell you is, make sure you listen to the audio BEFORE you accept the assignment. You can use some resources available on Express Scribe, such as volume control and increase/decrease speed to try to make it easier to understand and transcribe the content. However, sometimes it's just not worth it, because you won't be able to hear what is being said and you won't be able to provide a good transcription and translation service if you can't tell what people are saying...
Do you know of any software that could help improve audio quality?
Yes, there are programs that you can use to apply filters to audio files and try to minimize background noise or make voices sound more clear. Still, it will all depend on the original quality of the recording, and some filters may nevertheless result in garbled audio. In other words, the more you mess with a audio file, the higher the potential of I actually use one called Audacity, and I've taught a webinar about it, which was more geared towards offering voice over services, but has interesting information on how this piece of software can help you edit audio files.
Can you set a delay for time coding as you can do with subtitling?
You can definitely do that if you use a spreadsheet template and work on an equation that will automatically fill in a time code based on increments, such as 5-second time stamps. In other words, the spreadsheet would populate the time-code column instantly this way: 00:05, 00:10, 00:15, 00:20, 00:25, etc. However, that's only recommended for those projects when the client is not requesting a PRECISE time stamp. Considering the nature of the audio, someone could be talking for a while or there may be more than one voice in the recording, so it will not always be a "clean cut" time stamp that you could control based on a given interval.
Can Express Scribe plugins recognize different accents?
Some plugins may be available in different accents, so you'll see labels like "English (US)" and "English (UK)," for example, when you're installing an audio recognition plugin. Still, we may have a general idea of what certain accents sound like, but there's always a word here and there that may end up getting lost. Even inside the US, there are so many different variations in terms of accents, so you have to test the plugin if you get audio from, let's say, Boston or New Orleans, to see how the machine will interact with it.
Just as you can't trust machine translations blindly, you can't trust speech recognition either. If you use a plugin or another speech recognition solution, make sure you go through the transcribed content in its entirety to make it 100% accurate. Keep in mind you can't tell your clients "Oh, that's how the machine translation translated it" or "Oh, that's what the speech recognition plugin thought was being said" and you, as a human professional, must be the last quality assurance layer before a project is delivered.
Why do you recommend Google Docs for speech recognition? Can't Express Scribe do the job?
As mentioned above, Express Scribe comes pre-loaded with English speech recognition, and there are two accents available: US and UK. I personally never needed to upgrade to Express Scribe Pro, so you may have access to different audio recognition packages if you do an upgrade or purchase additional plugins.
However, if you're working with a different language, I suggest that you use the microphone function in Google Doc to recognize the audio and transcribe everything automatically. You do need to make sure your phone or tablet is configured to recognize speech in your language first―otherwise, it will transcribe gibberish in English by sound approximation, that is, what the foreign words sound like in English.
Once you're done with the automatic speech recognition performed by Google Doc, you can then use Express Scribe to add time stamps while you're reviewing the material, since you cannot trust this solution blindly, just as you would still review the output of a machine translation, for example. In a nutshell, using Google Docs for non-English speech recognition is just a short cut to save the time you would have otherwise spent typing everything out from scratch.
Can I use SRT files in Express Scribe?
SRT is a standard for subtitling projects. Considering that, SRTs already have text and time stamps, you cannot load it into Express Scribe. If you follow the process and transcribe a file in Express Scribe using your Word document template, you could―in theory―copy the timed transcription as plain text and save it as SRT. Express Scribe should also load video files, as you can see in their specifications, but I do recommend that you use subtitle-specific solutions to make sure you follow the correct subtitle formatting. I've presented two webinars focusing on subtitles: one about DotSub and one about YouTube.
What is encrypted dictation?
Let's say a transcriber is working for a medical office. Considering the nature of the material, the dictation file that would be sent by a doctor needs to be encrypted, so that personal and health information about the patient is kept safe. This is actually an advanced feature that Express Scribe offers, in that transcribers can connect to the client's server, using a username and password, to have access to the audio file and work on the transcription. The fact that the audio file is protected with a password means that this file transfer is encrypted.
Isn't transcription more expensive than translation?
If someone is a transcriptionist and transcribes for a living, they'll most likely charge more for it. However, when working with translation clients who need their audio transcribed AND translated, their frame of mind is that you'll be putting less effort in transcribing the recorded content ("You're just typing what you hear!") when compared to the actual work of translating the message into the target language. It all comes down to a value-added perception, so it's easier for you to charge "X" for transcription and "X times 2" for translation and make sure that the total price is covering the time you'll invest into the complete service.
Do you have an average rate for transcriptions?
You can calculate your transcription rate the same way you were able to figure out how much to charge for your translation work. In other words, once you've practiced enough―both with best-case and worst-case scenarios―you'll be able to have an average output, that is, how many minutes of audio you can transcribe during one hour of work. Having said that, if you charge X per hour and can transcribe 5 minutes of audio per hour of work, your per-minute rate would be X/5. So, just for the sake of making math easier, let's use an unrealistic high round number and say that you charge $1,000 per hour. Your rate when working with audio transcriptions would then be $200 per minute! ;-)