Our platform now allows podcasters to take advantage of Google's Speech-to-Text technology to produce automated transcripts.
How does it work ?
Machine learning technology applies neural network models to audio content to detect speech and convert it into text.
Why is it useful ?
Including a text version of audio content on web pages is a good SEO feature, allowing search engines to index more relevant keywords.
In addition, automated transcripts form a great start for a human-curated version, frequently only needing punctuation and correction of person and brand names.
How accurate is it ?
The accuracy of the transcripts depends on a lot of factors. Some key ones include:
- Maturity of model. Machine learning models for U.S. English has seen the most active development and produces the best results.
- Quality of content. Background noise and music, soft voices and inconsistent volume levels can drastically reduce the quality of results.
- Number of speakers. Different speakers and rates of speech can make it more difficult to extract useful transcripts.
How do i use it ?
Customers on our "Pro" or higher podcast packages can request access to this feature from their publisher admin portal.
After activation by our Ops team, episodes can be individually transcribed or it can be enabled on podcast channels to automatically transcribe all future episodes published to the channel. Publishers can choose to make transcripts available directly to end-users or only be visible in the admin portal.
All episode pages on our website automatically includes any transcript text as keywords in the HTML <meta> tag also.
How much does it cost ?
Automated transcription is invoiced on a per-minute basis.
When transcribing individual episodes the exact cost is indicated. Automated transcription of all new episodes published to a channel displays an estimated monthly costs, based on the past 30 days of activity.
For up-to-date pricing, please see our podcast pricing page or visit our podcast transcription FAQ.