simple_text_aggregator

Simple text aggregator for basic sentence-boundary text processing.

This module provides a straightforward text aggregator that accumulates text until it finds an end-of-sentence marker, making it suitable for basic TTS text processing scenarios.

class pipecat.utils.text.simple_text_aggregator.SimpleTextAggregator(**kwargs)

Bases: BaseTextAggregator

Simple text aggregator that accumulates text until sentence boundaries.

This aggregator provides basic functionality for accumulating text tokens and releasing them when an end-of-sentence marker is detected. It’s the most straightforward implementation of text aggregation for TTS processing.

__init__(**kwargs)

Initialize the simple text aggregator.

Creates an empty text buffer ready to begin accumulating text tokens.

Parameters:

**kwargs – Additional arguments passed to BaseTextAggregator (e.g. aggregation_type).

property text: Aggregation

Get the currently aggregated text.

Returns:

The text that has been accumulated in the buffer.

async aggregate(text: str) → AsyncIterator[Aggregation]

Aggregate text and yield completed aggregations.

In SENTENCE mode, processes the input text character-by-character. When sentence-ending punctuation is detected, it waits for non-whitespace lookahead before calling NLTK.

In TOKEN mode, yields the text immediately without buffering.

Parameters:

text – Text to aggregate.

Yields:

Aggregation objects (sentences in SENTENCE mode, tokens in TOKEN mode).

async flush() → Aggregation | None

Flush any remaining text in the buffer.

Returns any text remaining in the buffer. This is called at the end of a stream to ensure all text is processed. In TOKEN mode, returns None since tokens are yielded immediately.

Returns:

Any remaining text as a sentence, or None if buffer is empty or in TOKEN mode.

async handle_interruption()

Handle interruptions by clearing the text buffer.

Called when an interruption occurs in the processing pipeline, discarding any partially accumulated text.

async reset()

Clear the internally aggregated text.

Resets the aggregator to its initial empty state, discarding any accumulated text content.