Choosing between Automated & Human Transcription
Jul 02, 2024, Ushma KapadiaAs the world continues to shrink and linguistic boundaries continue to blur, Automated Transcription (AT) has become integral to business practices. Be it for market researchers/ journalists/ marketers/ lawyers or students; AT takes away the tedium and time associated with data analysis.
The term ‘Automated Transcription’ is typically used in context of software solutions that convert live/ recorded Audio or Video into Text. Think voice-to-text transcription appearing on a YouTube video you view. Think live transcription appearing in a webinar window you attend over Zoom. Think apps that transcribe podcast interviews you listen to. A human transcribing the same content could take hours, even days to accurately transcribe the entire interaction. Instead, Automated Transcription solutions do the same job in minutes.
On the other hand, ‘Human Transcription’ (HT) involves human intervention. Herein, native speakers are trained to fine-tune Tech-generated transcripts. Human transcribers understand human communication and can identify gaps or inaccuracies in content transcribed. In that sense, HT brings the best of Human and Artificial Intelligence (AI), to the transcription process.
The use of AT in Qualitative Research is gradually increasing. For an industry that heavily relied on Manual transcription, this is a significant change. Given Qualitative Research’s human-intensive nature, it is important to understand when and how using AT would be most beneficial to the research process.
The decision to use AT vs HT in Qualitative Research primarily hinges on 4 factors :
- Sector/business regarding which the research is being conducted.
Certain businesses involve use of hyper-specialised terminology and language. For instance, Qualitative Research in Pharmaceuticals involving generic drug names/ formulation names/ disease names/ symptom names. In such use-cases, there are two options:
- AT-only; where the study-specific terms are fed to the program; thus ensuring that human error in getting the terminologies right, is eliminated.
- AT double-checked by Human intervention; to maximize accuracy. - Project design and operations i.e. Sample size, Online vs Offline, Group Discussions vs In-depth Interviews, Recording equipment used, Researcher’s role during fieldwork. As a thumb-rule; small Online projects with interactions involving fewer voices (In-depth Interviews, Dyads, Triads) are better suited for AT.
Quality of Online recordings is usually superior to that of Offline recordings. Hence, generally speaking, Online recordings are better suited to AT.
Not all units of fieldwork are attended in-person, by the researcher. When a researcher attends in-person, more the likelihood that they are familiar with what was said, hence able to identify transcription inaccuracies and identify the context behind specific responses. Researchers present at fieldwork may use the transcript only to pick verbatims or double-check observations made during fieldwork. In this case, AT could suffice. On the other hand, researchers not present during fieldwork conducted in English might rely slightly more on the transcript. In such cases, HT could be a better choice. - Role of transcription in the research process
Where transcripts are a client-deliverable, HT is the recommended option. Doing so helps the researcher optimize time taken for accurate, high-quality, verified-by-human-eye transcription. On the other hand, if transcripts are meant to feed into the analysis process; AT can be the first choice. Human intervention can be applied for specific transcripts, at the researcher’s discretion. - Language in which the interactions are conducted
Generally speaking, AT is reasonably high (upto 90%), for popular global/ regional business languages eg. English-to-English, Spanish-to-Spanish.
In sum, here are broad pointers to help decide when to use what kind of transcription:
| Sector/business regarding which the research is being conducted | For hyper-specialised sectors – use HT, to maximise accuracy |
| Project design and operations | For small sample-sizes, one-to-one, Online interactions – use AT, to optimize time |
| Role of transcription in the research process | If transcripts are a client deliverable – use HT, to maximize accuracy & time |
| Language in which the interactions are conducted | For popular global/ regional languages – use AT |
At mytranscriptionplace, we specialise in high-quality Automatic Transcription and Human Transcription. Our HT process combines AI technology advancements and language expertise of native speakers; supplemented by a rigorous quality control process. All our transcribers are native and chosen after rigorous scrutiny. They use industry-standard AI-enabled tools to deliver timestamped, interactive, human-checked, peer-verified transcripts. Next, these transcripts are proof-read by our quality control team, for same-day delivery.
