Many podcasts hosts focus on their listeners, which is excellent. But, a part of the population cannot listen to a podcast. People hard of hearing or people learning a language use transcripts. Providing a transcript allows them to join a podcast host's audience.
To make a transcript, a podcast host has four options:
- Human Service
Each method has advantages and disadvantages. I used each method to create a transcript for the same episode.
I used the same criteria to evaluate which method is the best.
Speed is one factor used to evaluate the options. Speed is how fast the transcript could get developed. Next, I used accuracy, how many errors were in the transcript that I needed to fix. Finally, I used the price, how much money did it take to create the transcript.
Once a transcript gets made, a host has several options.
A transcript would increase the chances of finding the podcast on a Google search.
People will spend time reading the podcast transcript. This will increase the time on site. Which would make the podcast website more favorable with Google's search algorithm.
A host can take the transcript, add some formatting, and turn the episode transcript into a blog post. The blog post can get posted in different places for people to read it. The blog post can contain a link back to the podcast episode to promote the podcast.
A host can give their audience the ability to download the transcript. For the transcript to be downloadable, a host would have to put the transcript on a PDF. Then, create a button with the functionality for visitors to download the content. A host loses the opportunity to increase their domain authority. But the transcript can be downloadable and on the website at the same time.
Providing a transcript is extra work for a host. But there are benefits to providing a transcript.
When putting a transcript on the website, Google will index the page. Someone might search for some words used in the transcript. Then, the transcript page might pop up in the search results.
For example, someone might search "Open Podcast Community Podcast." Then, a page from the Building an Indie Business Podcast website might show up. Since in every episode, I mention the Open Podcast Community Podcast.
By providing a transcript, a host is expanding their potential audience. Hosts that provide transcripts have an advantage over those that do not. Those that need transcripts will not become fans of content they cannot consume.
The Speech-to-Text software that I used to transcribe the podcast is otter.ai. Otter has a free plan that gives users 600 free minutes a month. As well as 3 free audio files to get uploaded and transcribed, and each transcription can be 45 minutes long.
Otter can act as a backup for the audio. If anything goes wrong with the regular recording, a host can export the audio from Otter and use that.
A host can export the transcript from Otter with timestamps, like a monologue that ends up as one wall of text, and as a txt file. The premium service allows for unlimited minutes, uploads, and recording time. At the same time, the audio can get exported as a DOCX, PDF, and SRT file.
I used a 28-minute episode to test out transcripts. The episode was about launching a podcasting book on Product Hunt.
The episode transcript had 84 errors. most errors were "Product Hunt" and "um," which the software mistook as "product on" and "arm." I tend to have a problem saying "um" while recording. The software mistook that as "arm."
The transcript took 5 to 10 minutes to process and about 45 minutes to fix the errors. Creating and editing the transcript with Speech-to-Text took ~1 hour.
Otter, as a service, is a web app. If a host's internet goes down, the transcript gets lost. If a host hasn't used their three uploads, they could upload the audio file to Otter.
I uploaded one of the interviews that I have done. I noticed that the service has trouble with voices. The service could not tell the difference between the person I was interviewing and me. If a host does interviews, this is something that they need to edit themselves.
Turn on the Speech-to-Text software and then hit record. The Speech-to-Text software only records the speech. The clicks of the record button will not show up. If the record button gets hit and then the Speech-to-Text software gets turned on. The sounds will show up in the podcast episode. The transcript recording sound does not matter. This is because the text is the most critical factor for Speech-to-Text software.
The Product Hunt episode was available 12 hours after the episode got uploaded. The episode had 31 errors. Sometimes it would mistake Product Hunt for "product on." But many times, it understood it, which scared me.
Getting the transcript from YouTube has extra steps.
- Upload the video to YouTube
- Go to Channel Videos
- Then Details
- After, More Options
- There will be "English" when it's ready
- Click the 3 dots on "English"
- Edit in Classic Studio
- Download as SRT
- Take the SRT file to another website
- Convert SRT to TXT
- Fix errors
While the YouTube transcript is free, a host depends heavily on a massive company. YouTube has many videos to transcribe. There's no way of knowing how long the transcript will take to get completed. At the same time, they might not even think transcribing a video is worth it. This happened with a video that I produced that was 2 minutes long. I never got the video's transcript.
The YouTube transcript is accurate and accessible. Although, hosts do not know when they will get the YouTube transcript. This means that a YouTube transcript is unreliable. A host might record a last-minute episode. They might not be able to provide a transcript for the episode with YouTube.
A podcast host can pay a transcription service for a human to write the transcript for them.
The price of the transcript depends on how long it is and when a host wants it. For a 5 day delivery, they charge $0.85/minute, $1/minute for 3-day delivery, and $1.20/minute for 1-day delivery. The Product Hunt episode would cost a minimum of $23.80 and a max of $33.60.
The service offered me a free trial for the first 5 minutes. I sent them the two-minute podcast episode that I produced. After 3 days, I sent the Product Hunt episode. I never got either transcript. Being that I never got the transcripts, I cannot speak to the quality of the work.
In terms of the options that the service allows you to choose is a clean or full transcript. This means that a host can choose a transcript without all the errors, such as "um." The full transcript has all the errors. They can add edits such as timestamps or speaker identification too.
There's always the option for a podcast host to transcribe the episodes themselves. This can be a time-consuming task if the podcast is long.
For the experiment of writing my own transcript, I wrote 1 minute of the podcast episode. To write that one minute, it took 10 minutes. That means that a 30-minute podcast episode would take me 5 hours to transcribe fully. Spending 5 hours on a transcript is not ideal.
The advantage of transcribing the episode myself is that there were no mistakes. I can style the transcript in any way.
The best option for transcribing is to use Speech-to-Text software. Once the episode gets processed, A host will still need to edit the mistakes. Speech-to-Text software will produce more errors compared to YouTube or human service. The errors will take a shorter amount of time to fix from Speech-to-Text. It would take more time to get a transcript from the other two options. The same principles apply for writing a transcript from scratch. The Speech-to-Text software is quick, reliable, and a cheap option.
Some podcast hosts make their audience pay for transcripts; this is not inclusive. Paying for transcripts is exclusive because the podcast is free. Yet they make people pay for reading the transcripts. Which is not right.