Finally! I’ve been wanting this for the last two years, ever since I started podcasting. Google Docs has been phenomenally helpful as an online resource for everything I do podcasting wise.
After using Google’s speech-to-text engine (or voice-to-text, or voice recognition, depending on who you ask) in Google searches with great success, I’ve been wanting it to come to Docs.
It’s here!
And now we can use their amazingly accurate voice dictation to transcribe our podcasts in real time.
Now, before you get too excited, here are some things you need to know:
- Speech to Text will always be flawed. There are just too many variations in the way we say things, so mistakes are pretty much unavoidable. However, Google does a great job of matching content with context. It will actually correct words after it reads the sentence! So a sentence like, “I when to the store” will turn into “I went to the store”.
- A cough can be mistranslated into a word. In fact, any background noise can be picked up and made into a word. Hopefully, our studios are silent and our coughs are minimal.
- You will spend a lot of time correcting and modifying the text when it’s done, and it’s ultra important that you do so! Otherwise, when you said, “Grandma fried some corn” but it got translated into “Grandma died from corn” (which is grammatically correct), you might be embarrassed when someone points it out to you.
Otherwise, this is a pretty darn cool tool to have as a podcaster. I hope Google turns this into a learning tool, where it can learn the way we talk over time. Maybe someday.
Bonus Tip: There’s no reason you can’t take a current interview and plug it into your microphone port and have it translated. Go ahead… try it!
UPDATE: I can now take any voice recording and send it to Google Docs and use Voice Typing to transcribe. I do this by playing the audio from my computer, which sends the sound to my mixer, which in turn sends the sound back into my computer through the AUX SEND port.
For a really basic interpretation of what I just said, just think of it as plugging an audio cable from the headphone jack (OUT) of your laptop to the microphone jack (IN). It creates the same loop either way, but I have a lot more control through the mixer obviously.
Enjoy really accurate voice transcription, and have fun editing!
Excellent discovery! Thanks for sharing.
You’re welcome! I’ve been using it for the last two episodes and it is really accurate. I wish it would guess at punctuation, but for the most part, it works very well.
The only downside is that sometimes it just stops for no reason. I’m hoping it’s just because it’s buggy and they’re still beta testing it.
Regardless, I can see this tool growing into something very good.