The four challenges of transcribing audio material

The four challenges of transcribing audio material

Wordpar International can transcribe audio content in a wide variety of languages ​​using both standard and proprietary customer tools. There are many reasons why businesses turn to transcription services. Some use international transcriptions to build voice recognition databases, others to analyze call histories or to provide SEO data for localized video content. The majority of our transcription jobs involve transcribing call center conversations between customers and employees.

For market leaders, call center conversations can contain a wealth of information that allows deep insights into what is happening in their company. Every conversation with a customer is important and can contain valuable information about market trends, problems, employee performance, emotions of the customer or the company's products and processes.

Analysis of these conversations can be the ultimate source of information about customers. Between the greeting and the farewell, there are thousands of words and phrases in the customer-employee conversations that have been collected, which can contain important data about how both customers and call center agents are positioned.

In an analyzed form, this information can contribute significantly to an increase in company performance and a better customer experience. The four main challenges involved in transcribing audio material will be discussed here.

First of all, however, the following should be clarified:

• Why does the customer need the transcription service?

• What are his current challenges?

• What output do the customer's software engineers need?

Once these points are clarified and a schedule is established, a team of reliable transcriptors must be in place so that the customer's deadlines and quality expectations can be met. That alone can be a real challenge.

First challenge: selection and long-term commitment of the team members

Depending on the language being transcribed and delivery times, it can be difficult to find local transcription and quality assurance resources that can stay on board for the duration of the project. The majority of our transcription projects require 25 to 30 dedicated employees in order to meet our customers' delivery deadlines.

The teams are composed as follows for each language:

• Resource manager who finds, selects and interviews suitable employees for the project

• Transcriptors for transcribing the audio content

• Coordinating language and QA specialists who carry out the necessary quality assurance reviews

• Senior project manager who ensures that the project runs smoothly

Producing high quality, professional transcriptions can take a lot of patience and effort. Depending on the type of transcription required, the work for the transcriptor can be tedious and even frustrating. And that is one of the main reasons why some drop out in the course of the project, which in turn means that the search for suitable transcripters and their training can drag on throughout the project.

Also, depending on a few factors that we will discuss below, it can take anywhere from four to ten hours or more to transcribe an hour of audio material.

Second challenge: quality of the audio material

Here are a few factors that determine the difficulty of transcribing audio material:

• Quality of the audio material

• Whether the conversation follows a script (usually marketing content) or is conducted freely (e.g. call center discussions)

• Content of the audio material

• Voice quality of the speakers

• Different dialects and accents

• Background noise

Most of the files that we receive for transcription are free conversations between customers and call center agents, so that the quality is sometimes poor and only a certain percentage of the material can be used.

Since the audio files contain mostly natural conversations, we encounter the following problems:

• Simultaneous speaking

• Wrongly pronounced words

• Jargon / technical language

• Incomprehensible words

• Inaudible speech due to too much background noise

To meet this challenge, we have set up two teams for such projects: an evaluation team and a quality assurance team.

Evaluation team

• Unusable material is documented and brought to the attention of the customer

• New files are requested and evaluated

Using this team has the advantage that reliable audio material is always available. Because trying again and again to decode poor quality audio material would significantly delay the work of the transcriptors and thus reduce throughput.

Quality assurance team

This team is responsible for checking every single transcribed file before it goes back to the customer. This ensures that the established rules have been followed and that the quality of the transcription files meets the customer's expectations.

• The work results of the transcriptors are validated daily with the help of our testing tools and the knowledge gained is regularly shared with the teams.

• On this basis, poorly performing transcriptors (in terms of quality and throughput) are either retrained to meet our schedule and quality expectations, or replaced with new employees.

Third challenge: training

Since most of our transcripters have no experience with clients' transcription and validation tools (although they may be able to transcribe using standard industry tools), we need training plans and support documentation for each project.

Part of the transcription training is for example:

• Structure of breakpoints / natural separation points in the conversation (the maximum length is specified by the customer's engineers)

• Marking different voices in conversation

• Dealing with regionally different vocabulary and colloquial language

• Tagging of incomprehensible parts of the conversation or parts with a strong accent

• Explanation of the rules for acronyms and abbreviations

• Tagging of exclamations, hesitation, pauses in conversation, etc.

• Learn the keyboard shortcuts and functions for the spell checker and dictionary

• Detecting Known Bugs

Since the various transcription tools may be unknown to even experienced transcripters, a great deal of time is devoted to familiarizing teams with how the tools work and their limitations.

In addition, it is often the case that the customer specifies very precise rules for the files, the correct implementation of which takes a long time in the first few moves of the project. Daily QA checks and reports from our own QA team on the work of the transcriptors help to continuously improve and optimize progress. The teams in all projects can exchange important information in shared questionnaires and Skype groups.

Fourth challenge: security

Due to the sensitive content of call center calls, which may contain the customer's e-mail address, address, telephone number, credit card details, etc., a high level of security must be ensured.

Audio material and transcribed content must be stored in a secure network and a strict confidentiality agreement is concluded with each project member involved.

• SFTP folders are created and password-protected access areas are set up.

• No content can leave the network without authorization.

• Team members cannot download files, which can be problematic depending on where the transcriptors are located in their home countries. Hubs need to be set up in optimal locations to give teams uninterrupted access.

• After completion of the project, all original and edited content will be deleted from our servers and their deletion will be certified.

Conclusion

While transcription projects can be challenging, working closely with our clients, joining their team to understand the specific requirements, has always helped set the process up for such

 

Strengthen and optimize projects.

The Wordpar International transcription service team in Bangalore has developed a professional solution for the implementation of high-quality transcription projects within agreed delivery times.

Would you like to find out more about our transcription service?