Speaker ID

Case Study: Atlanta Market Research Company Transcription

Case Study: Atlanta Market Research Company Transcription 4218 2802 Atomic Scribe

The Client

The client was a respected market research company in the Atlanta, Georgia area who required transcripts for eight focus group audio recordings, with five to 10 speakers in each recording.

The Challenge

Our biggest challenge was successfully transcribing and identifying as much speech as possible. The groups all had multiple participants who were young mothers from the same area (and, thus, all with similar accents).

Other challenges included:

  • Southern accents
  • Unfamiliar jargon
  • Broken English
  • No video to help identify speakers
  • Multiple speakers talking at once

The client also needed groups back within four days, which was a tight deadline for the large amount of recordings. As these were multiple-speaker focus groups with difficult audio, the project was not eligible for our machine transcription option and needed to instead be transcribed in-house by our workers.

“Atomic Scribe was a joy to work with. Kathryn kept me updated on progress so that I didn’t have to worry about files being sent back late. Instead, everything was well on time and extremely high quality. 100% value for the money!”

– The Client

“Atomic Scribe was a joy to work with. Kathryn kept me updated on progress so that I didn’t have to worry about files being sent back late. Instead, everything was well on time and extremely high quality. 100% value for the money!”

– The Client

The Atomic Scribe Solution

An Atomic Scribe project manager immediately reached out to the company via the client’s preferred communication method (e-mail) to confirm the project details and provide an estimate for the project cost. We then had the client create a free account through our dedicated client portal so they could upload their files to our server.

Once we had the recordings, we listened to each recording and assigned files to our transcriptionists based on which groups suited each transcriptionist best. When the first transcriptionist completed a file, a proofreader then listened to the entire file again to edit the transcript, if necessary. Our two-tiered approach ensured maximum quality for the client.

The Result

Because of Atomic Scribe’s efficient system and communication, we finished the project within the requested timeframe. Atomic Scribe also:

  • Adhered to a client-specific template
  • Achieved 98% accuracy on clear-audio files and averaged 94% accuracy on difficult files
  • Successfully identified all respondents to the best of our ability
  • Allowed for easy upload and download of files through the client portal
  • Provided a simple way for the client to pay their invoice through the client portal

All in all, the client rated our service five out of five stars, with accuracy and communication as the top two listed services provided by Atomic Scribe. We have continued to work with this client on multiple market research transcription and translation projects.

Rev vs. Atomic Scribe Transcription

Rev vs. Atomic Scribe Transcription 2500 1667 Atomic Scribe

12/2/2019 Edit: This post was written in 2016, but we believe these points are still relevant today. As such, this post has not been altered below this point. 

There are a lot of transcription companies out there, so how do you choose the one that’s best for you? In this piece we’ll lay out some key differences between Atomic Scribe and Rev, a popular company that we used for four focus groups to test out their transcription quality. Rev boasts prices of $1/audio min, no matter the audio quality or length. Here’s what we found.

Rev Has No Real Delivery Deadline

Although it is stated on Rev’s site that orders are generally delivered in 24 hours, it actually depends on the file. We also found that they did not give us status updates on a file’s progress in a timely fashion, which can be extremely worrying for those with a deadline.

For example, one of our focus groups files was ordered on Jan. 31st and labeled by Rev as due on Feb. 3rd (already well past 24 hours)… but it wasn’t delivered until three days later on Feb. 6th. That’s a whole week from the initial order date, which is longer than our standard turnaround time. No discount or refund was offered.

At Atomic Scribe, deadlines are important. We have three tiers (rush, standard and discount), and we adhere to whichever tier the client chooses. But sometimes files are harder than expected and more time is needed, in which case we always contact the client immediately and ask if this is acceptable. We also provide a discount if a deadline extension is necessary.

Speaker ID Inaccuracy

Limited speaker ID is included in Rev’s low price, but unfortunately most IDs were unhelpful. While some speakers were correctly identified as “Male” or “Female”, other times names were used as labels sporadically. Basically, there was no uniformity.

In one transcript the speakers were labeled Speaker 1, Speaker 2, Speaker 3, etc. However, it was clear from a re-listen that the transcriptionist could not actually tell the speakers apart and that many labels were incorrect. A good example of this is the number of identified speakers went up to 6 when there was actually 8 participants, but no Speaker 7 or Speaker 8 were found in the transcript (yet they were found by our own transcriptionists).

At Atomic Scribe, we see that as a waste of time. Yes, we charge more for speaker ID, but with that extra charge we come as close to 100% accuracy as we can get. There’s no point in having speaker ID at all if it isn’t accurate, and there should always be the option to add it to an order if it’s helpful to the client. It should also be uniform and not a different format in every transcript.

Too Many Inaudibles

One transcript (the one not delivered until a week after ordering) actually had 337 inaudibles! When our transcriptionist finished it on a re-listen, there was—no joke—10 inaudibles left. There were also quite a few inaudibles across the transcripts that were easily found by just doing research, such as a town or company name.

We’ll admit our audio wasn’t perfect and focus groups are hard, but that’s why most companies charge more for multi-speaker files and pay their transcriptionists more for them. A flat fee system like Rev uses treats all files the same, which hurts transcript quality.

  1. No real delivery timeline

  2. Wrong speaker ID

  3. Inaudibles

  4. Unhappy workers

  5. Poor quality

Low Pay Scale

After we received the transcripts from Rev, we were curious about their transcriptionists. First off, U.S. citizenship is not a requirement to work for Rev. That could mean a larger amount of foreign workers whose native language isn’t English.

We also found out that Rev pays their transcriptionists on average less than half what we pay ours (according to their own site), which probably accounts for the inaudibles. After all, if a worker isn’t receiving a fair wage, what incentive do they have to turn in a good transcript? If their contract is terminated, there are plenty of other companies that pay the same wages as Rev (and many that pay better).

Lower Quality

The transcripts widely ranged in quality. Some transcripts were acceptable, but some had to be almost completely redone. And that’s what is reasonable to expect from a low-cost service, as is expected in every industry. You know you will never receive the best quality, but you hope and pray that you won’t receive the worst either.

That might be okay for some of your projects. But for a higher price (that you only pay once), you get peace of mind, accuracy, correct speaker ID, uniform formatting, communication, research of names and companies, and skilled transcriptionists. It seems to us an easy choice to make.

Note: We confirm that we paid full price for Rev’s services for files in 2014 and are not affiliated with the company. This is a truthful review with no attempt to distort facts.