Skip to content

Determining the Most Accurate Transcription Services

Posted in Student Blog Series

Featured Image Provided by Content Tech

In today’s world where technology aids human beings in a variety of realms, advancements within transcription services and tools have been growing in importance. In this blog post, I will provide an analysis of different transcription services while listing out the advantages and disadvantages of each.

Introduction to Transcription

Transcription is essentially a procedure performed to record any live speech, video, or audio and convert it into a written form [6]. This written form can serve a multitude of purposes. These transcription services are used to transcribe seminars, interviews, lectures, meetings, and more [6]. 

Benefits

Transcription provides numerous benefits for a variety of people. Transcripts can help improve turnaround times, meaning that whenever audio or video materials play a major role, transcripts can aid people in speeding up their workflow [7]. The transcript, as a written document, can allow people to pinpoint the sections they wish to work on, instead of returning to the audio/video materials to tediously find the section [7]. Productivity is significantly improved when people do not have to switch back and forth between audio/video materials and their other notes or written materials [7]. With a transcript, they can easily complete their tasks without having to switch between different mediums.

Another benefit of transcription is that the value of the content increases [7]. In order to make video/audio content searchable, businesses need to transcribe or caption it so that humans and bots can parse the transcriptions to better understand what type of content is contained within the video/audio [7]. By making use of transcription, content can easily be searched and found. 

An additional benefit of transcription is that it increases accessibility [7]. Any public audio and visual materials need to be accessible to all viewers, including anyone who has a disability [7]. By transcribing and captioning this kind of content, every potential viewer is able to have access to these materials in at least one form–written, visual, or audible [7]. 

Manual Transcription Tools

Transcription can be done manually and there are tools, like the ones presented below, which can help make the process easier. 

OTranscribe

OTranscribe is a free, open-source tool that allows users to transcribe straight from a browser [1]. To begin the process, a user can upload their audio/video file or enter a link to a YouTube video [1]. Users need to manually transcribe themselves as it does not automatically convert audio into text [1]. OTranscribe offers a simple user interface where the left half allows the users to control the audio player and on the right half, the user can view and edit in the text editor [1]. It provides features like interactive timestamps and automatic saving capabilities [1].

GoTranscript

GoTranscript is a 100% human-generated transcription service that has starting charges from $0.72 per minute, 99% accuracy, and 6 hours turnaround time [5]. It offers transcription, translation, captioning, and subtitling services [5]. Their team follows a four-step process in order to maximize accuracy [5]. The first step is transcription where files are divided into sections and transcribed by professional transcriptionists [5]. The second step is review where all the sections are edited, and timestamps and speaker tracking are added [5]. The third step is proofreading where all sections are merged and proofread to make sure that the text is consistent [5]. The last step is quality check where the transcript is proofread another time to ensure over 99% accuracy [5]. It provides a free trial where a user can have five minutes transcribed.

Automatic Transcription Tools

While transcription can be performed manually, the following tools can help expedite the process.

Temi

Temi is an AI-based transcription service that uses a proprietary algorithm in order to transcribe audio and video files [2]. It provides a free trial through which an individual can gain access to all features to generate one transcript under 45 minutes [2]. After that, it costs $0.25 per audio minute and there are no additional costs [2]. The turnaround time depends on the length of the files, but it claims to provide transcripts back in minutes [2]. Some of its features include a simple editing tool where users can quickly clean-up the provided transcript, review and edit where a user can adjust the playback speed and easily skip around, custom timestamps where Temi knows the timing of every word and allows the user to add any timestamp, speaker identification where it marks the change of every speaker and labels them, and export transcript where a user can download the transcript into text (MS Word, PDF) or closed caption files (SRT, VTT) [2]. 

Otter

Otter is a transcription tool that helps users generate notes for meetings, interviews, lectures, and other important voice conversations through the use of an AI-powered assistant [8]. They provide different plans with a variety of features. In the Basic plan, which is free, the user can record and playback, transcribe in real time, make edits, search by keywords, share with collaborators, export a text file, transcribe a maximum of 600 minutes per month, and transcribe a maximum of 40 minutes per conversation [8]. The limitation of this plan is that users cannot upload their own audio and video files, export in different formats like PDF, DOCX, and SRT, and upload a custom vocabulary, which are all features that are provided in the Pro plan that costs $8.33 per month [8].

Manual and Automatic Transcription Tools

Some transcription tools, like the ones described below, incorporate AI to transcribe files and humans who manually transcribe files and/or review the AI generated transcripts. 

TranscribeMe

TranscribeMe uses both transcriptionists and AI to create transcripts based on files that users upload [9]. It provides transcription, speech recognition, translation, and data annotation services [9]. There are different types of transcription plans, but the most basic one (First Draft plan) involves human/manual transcription, costs $0.79 per minute, has 98% accuracy, delivers within one business day, and provides speaker IDs and timestamps as well [9]. The automatic transcription involves speech recognition costs $0.07 per minute [9]. 

Rev

Rev is a transcription service that also uses professional transcriptionists and AI to generate transcripts based on a user’s uploaded files [3]. Rev provides one free trial for 45 minutes [3]. It offers different plans starting at $0.25 per minute, 80% accuracy, and a 5 minute turnaround time [3]. The process begins with the user uploading their files, then the user can review and edit, and finally, they can share and export [3].  

Transcribe

Transcribe is another free tool that offers users different options to generate transcriptions [1]. Users can convert audio to text with automatic transcription, convert speech to text through fast and accurate dictation, and slow down audio and transcribe with tightly integrated player and editor [1]. Transcribe provides 30 minutes of free automatic or self transcription. If services are to be used after the free trial terminates, then the self transcription costs $20 per year, whereas the automatic transcription costs $20 per year plus $6 per audio hour. Some features include creating video subtitles with export formats in SRT and VTT files and built-in automatic text expander where users can define acronyms for any frequently used words or phrases [1]. 

Analysis of Transcription Tools

To test all serven of these transcription tools, I recorded an audio clip of myself narrating A Little Cloud by James Joyce [4]. After recording, I uploaded the file to each of the transcription services and compared all of the transcripts using a difference checker against the text of the original story. I chose to upload the first five minutes of the recording only as one of the tools’ free trials allows a transcription of an audio clip that is five minutes in length. In order to ensure that all comparisons are fair and equal, I only had the first five minutes of the audio clip transcribed. Since OTranscribe requires manual transcription, I have omitted it from this analysis. After completing this process, the accuracies of the remaining seven transcription services were revealed (Figure 1). 

Starting with GoTranscript, the transcript it produced had 21 punctuation errors, 3 capitalization errors, and 3 incorrect words. Additionally, I evaluated the transcript from Temi which made 52 punctuation errors, 10 capitalization errors, and 28 incorrect words. For Otter’s transcript, I found that there were 59 punctuation errors, 13 capitalization errors, and 26 incorrect words. After that, I analyzed TranscribeMe’s manual/human transcription through the First Draft plan which made 40 punctuation errors, 10 capitalization errors, and 22 incorrect words. Then, I looked at TranscribeMe’s automatic transcription through speech recognition which created 52 punctuation errors, 12 capitalization errors, and 22 incorrect words. Next, I analyzed Rev’s generated transcript and that created 52 punctuation errors, 12 capitalization errors, and 29 incorrect words. After checking Transcribe, I found that it generated a transcript with 56 punctuation errors, 18 capitalization errors, and 21 incorrect words. There were many errors that occurred through all the transcription services. 

Transcription ToolTranscription TypeCost per minutePunctuation ErrorsCapitalization ErrorsIncorrect WordsTotal Errors
GoTranscriptHuman/Manual$0.72213327
TemiAutomatic$0.2552102890
OtterAutomatic$0.00 591326108
TranscribeMeHuman/Manual$0.7940102272
TranscribeMeAutomatic$0.0752122286
RevAutomatic$0.2552122993
TranscribeAutomatic$0.10 + $20/year56182195
Figure 1. Summary of Errors Generated by Transcription Tools

Main Takeaways

Out of the seven transcription services that were analyzed, I determined which one would be the most accurate depending on the number of total errors produced by the transcripts. Based on the analysis of the transcripts, it seems that GoTranscript is the most accurate out of these seven transcription services because it had the fewest total errors, 27 errors. 

Regardless, it is important to understand these transcription services in terms of their costs as well. Only OTranscribe is completely free; however, it requires manual transcription while offering a usable interface. Among the transcription services which require payment, the cheapest option seems to be TranscribeMe which charges $0.07 per minute. 

Overall, transcription services are incredibly helpful when one wants to convert any audio or video file into a text format. After analyzing several different transcription services, it is evident that some are more accurate than others, but have different costs associated with them. If accuracy is the most important aspect of the transcript, then one should explore GoTranscript. If cost is an important factor, then one should consider TranscribeMe. If accuracy and cost are both equally important parts, then TranscribeMe is the best option with a total of 86 errors.

Works Cited

[1] 3 Free Transcription Tools You Can Try – Ai-Media. (2020, November 12). Retrieved November 22, 2020, from https://www.ai-media.tv/the-best-free-transcription-software

[2] Audio to Text Automatic Transcription Service & App. (n.d.). Retrieved November 22, 2020, from https://www.temi.com/

[3] Automated Transcription: Speech Recognition, Voice to Text Software. (n.d.). Retrieved November 22, 2020, from https://www.rev.com/automated-transcription

[4] Classic short stories. (n.d.). Retrieved November 23, 2020, from https://www.world-english.org/stories.htm

[5] Go Transcript Transcription Services. (n.d.). Retrieved November 22, 2020, from https://gotranscript.com/

[6] Kanabar, G., Asthana, R., Trends, S., Carere, C., Singh, A., Cherisien, W., . . . Khan, A. (2018, January 29). An Introduction to Transcription Services. Retrieved November 22, 2020, from https://socialnomics.net/2017/12/14/an-introduction-to-transcription-services/

[7] Myers, E., & Myers, E. (2020, July 06). Why Transcribe? 4 Ways Transcription Benefits Your Workflow — Rev.com. Retrieved November 22, 2020, from https://www.temi.com/blog/transcription-benefits-your-workflow/

[8] Otter Voice Meeting Notes. (n.d.). Retrieved November 23, 2020, from https://otter.ai/login

[9] TranscribeMe. (n.d.). TranscribeMe: The most accurate transcription starting at $0.79 per minute. Retrieved November 22, 2020, from https://www.transcribeme.com/

[10] Transcription Services. (2019, November 04). Retrieved November 22, 2020, from https://contentech.com/services/transcription-services/

Hariti outside in the winter wearing a winter coat
+ posts