AITranscriberNodeEditor Documentation

Overview

The AITranscriberNodeEditor is a component within the Vantage analytics and data platform that enables users to configure an AI-driven audio transcription node. This node is designed to automatically transcribe audio files from specified sources, such as file URLs or direct uploads. The transcribed text is then returned as a new column within the dataset, allowing for powerful analysis and integration into further workflows.

Purpose

The principal purpose of the AITranscriberNodeEditor is to facilitate the easy configuration of audio transcription tasks. It allows users to specify the audio input, the desired output format, and the efficiency of processing through batch size settings. This component streamlines the process of converting audio data into text, which can then be utilized for subsequent analytics or machine learning tasks.

Settings

Audio Column

1. Name: Audio Column
2. Input Type: Dropdown / String
3. Description: Designates the column in the upstream dataset that contains the URLs to the audio files (e.g., MP3, WAV, M4A) to be processed. The component will pull audio from this column for transcription.
   • Effect of Change: If altered, the AITranscriber will use the new audio column value, changing the source of the transcription.
4. Default Value: audio_url

Output Column

1. Name: Output Column
2. Input Type: String
3. Description: Specifies the name for the new column in the dataset where the transcribed text will be stored.
   • Effect of Change: Changing this value will modify the name of the column that receives the transcript output, affecting how users can reference the transcribed data downstream.
4. Default Value: transcript

Batch Size

1. Name: Batch Size
2. Input Type: Numeric (Range Slider)
3. Description: Defines the number of audio files to be processed in parallel during transcription. A higher batch size can improve transcription speed but may also increase load on external APIs and may incur additional costs due to higher usage.
   • Effect of Change: Adjusting the batch size will influence how quickly audio files are transcribed and can affect system resource utilization.
4. Default Value: 5 (range between 1 and 10)

Data Expectations

The AITranscriberNodeEditor expects two primary types of inputs:

1. Audio Source:

   • From URLs: If the user specifies a column with audio file URLs, those URLs must point to accessible audio files in supported formats (MP3, WAV, M4A, WebM, OGG, FLAC) not exceeding 25MB.
   • From Direct Uploads: If a file upload node is connected upstream, the transcriber will directly use files uploaded, negating the need for audio URLs.

2. Output Destination:

   • The name specified in the "Output Column" field must be unique in the dataset to avoid overwriting other columns.

AI Integrations

The AITranscriberNode integrates with external AI transcription services. When audio files are processed, requests are sent to these services which return the transcribed text. The efficiency of this process can be influenced by the chosen batch size, which sets the number of concurrent transcription requests.

Billing Impact

Depending on the external AI transcription service used, processing a larger number of files through increased batch sizes may lead to higher API usage costs. Users should monitor their usage to manage billing effectively, especially when processing large datasets or frequent transcription requests.

Use Cases & Examples

Use Cases

1. Content Creation: A media company can use the AITranscriberNode to transcribe interviews and podcasts automatically, enabling rapid content generation for articles and social media posts.
2. Legal Transcriptions: Law firms may utilize the node to transcribe recorded depositions and testimonies for review and analysis, saving time and minimizing manual transcription errors.
3. Accessibility Compliance: Educational institutions can implement the transcription of audio lectures to provide closed captions and written materials for students with disabilities, ensuring compliance with accessibility standards.

Example Configuration

Use Case: Transcribing Weekly Podcast Episodes

Configuration Data:

{
    "audioColumn": "podcast_audio_url",
    "outputColumn": "podcast_transcript",
    "batchSize": 5
}

Description: In this scenario, the AITranscriberNodeEditor is configured to take entries from the podcast_audio_url column, which contains the URLs of the weekly podcast episodes. The resulting transcriptions of these episodes will then be stored in a new column named podcast_transcript. By setting a batch size of 5, the system will process up to five audio files simultaneously, optimizing the transcription time while balancing API usage costs.