Speech to text - Contexta
Accurate transcripts of phone conversations
Quickly receive transcripts of audio files.
A good understanding of why people contact your company is extremely important. With these insights you can improve your communication, reduce churn and cut operational costs significantly. With the Contexta API you will receive accurate transcripts of your phone calls to achieve this.
The transcription module is tailored to phone speech to ensure optimal accuracy.
Speech recognition is not a one-size-fits all solution. Customize this module for it to (better) understand the relevant t words for your industry.
You will also receive meta data on timestamps, confidence scores and speaker id's. Speaker diarisation works on both stereo and mono files.
How it works
Contexta offers an accurate API for transcribing telephone data (English and Dutch). Although the general models already have a high accuracy, customizing models for optimal accuracy can be done easily. Audio is uploaded per API call and is transcribed in less than 50% of the audio time. Contexta transcripts come with meta data (time-stamps, speaker-id and confidence).
|Minutes per year||Pricing per minute|
|< 50.000||€ 0,104|
|50.000 - 100.000||€ 0,091|
|100.000 - 500.000||€ 0,078|
|500.000 - 1.000.000||€ 0,065|
|1.000.000 - 2.000.000||€ 0,052|
|2.000.000 - 3.000.000||€ 0.039|
|> 3.000.000||€ 0,026|
After signing up, you can start testing the speech to text engine. This is available for you to experience the possibilities.
If you like it, apply from your dashboard to take this API into production.
3. Go live!
Now you can start using the API in your own application and transcribe your conversations!
Labelling of phone calls is often inaccurate, inconsistent and too generalized. With an accurate transcript of a phone call you can automatically extract the following elements per conversation: duration / amount of silence per phone call, important topics, key words, sentiment, agent / customer speaker distribution.
Detect phrases for compliancy purposes
Call Center agents are often required to manually annotate compliance phrases. Checks on date of birth or a clear mention of terms and conditions are examples of things that need to be mentioned during a sales related phone call. With transcripts and text-mining this process can be automated, eliminating the costly, inefficient and error prone manual process.
Monitor quality of customer service
Monitoring your customer service agents is important to ensure high quality. Currently employee quality is often monitored by randomly listening to recorded phone calls. This way quality managers often listen to calls without relevant topics and learning elements. With transcripts it’s easy to filter and extract calls that matter.