Is there an app to have an LLM voice conversation using standard text-to-speech?

Exploring Voice-Only Interactions with Large Language Models: Is There a Pure Text-to-Speech Solution?

In recent years, large language models (LLMs) like ChatGPT have revolutionized the way we interact with AI, enabling natural language conversations via text. However, as these models increasingly integrate voice functionalities—whether through built-in features or third-party applications—users have noted some limitations in how these voices sound and behave.

The Challenge: Expressive Voices with Unwanted Fillers

One common observation is that voice outputs generated by ChatGPT and similar models tend to incorporate emotional intonations and filler words such as “um,” “er,” or “so.” While these nuances can make conversations feel more natural and human-like, they can also detract from certain use cases where a neutral, robotic tone is preferable—such as in professional settings, tutorials, or accessibility tools.

Many users have expressed a desire for a “robot mode”—a way to listen to LLM responses that are read out in a straightforward, emotionless manner, akin to standard text-to-speech (TTS) systems. This would enable more consistent and predictable vocal interactions, especially if the goal is to read out formatted information or to avoid unintended conversational nuances.

Current Solutions and Limitations

Most TTS engines and voice synthesis systems inherently produce speech that is relatively neutral and devoid of fillers. Some AI chatbots integrate TTS modules, but these often inherit the same emotional prosody and filler tendencies as the underlying models, resulting in speech that sounds overly expressive or “human” in a way that may not always be desirable.

Furthermore, popular voice assistant platforms and AI chat integrations may not offer granular control over voice intonations or filler words, making it difficult for users to customize the speech output to meet their specific needs.

Is There a Way to Achieve a Pure “Robot” Voice with LLMs?

Currently, there is no widely-known, dedicated application that allows users to engage in voice conversations with LLMs while maintaining a strictly mechanical, emotionless tone solely through standard TTS. However, some approaches may help approximate this:

• Custom TTS Models: Some advanced TTS systems enable fine-tuning to produce more robotic or neutral speech. For example, companies like Microsoft, Google, and startups specializing in voice synthesis offer customizable voice models that can be tailored to reduce emotional expressions.

• Third-Party Voice Conversion Tools: Certain voice modulation tools can