Gandr TTS as a LangChain tool.
One pip install, then a standard agent tool your LLM can call.
Ask Claude Code
Ask ChatGPTThe whole integration
Install, construct, hand to the agent.
pip install gandr-langchain
from gandr_langchain import GandrText2SpeechTool
tool = GandrText2SpeechTool(voice="gandr-leo")
The tool is a standard LangChain tool, so it plugs into any agent that accepts a tool list. The key comes from GANDR_API_KEY.
What it needs
What it takes
| One package | pip install gandr-langchain |
| A key | GANDR_API_KEY in the environment |
| LangChain | Any recent version with tool support |
| Python | 3.9 or newer |
Speech as a tool your LLM can call
A LangChain agent already picks tools by description. GandrText2SpeechTool is a standard LangChain tool, so the agent decides when an answer should be spoken and calls the tool like any other.
The voice is set at construction, so an agent can carry a consistent voice across turns. The key comes from GANDR_API_KEY in the environment.
What the flat stream changes
An agent that speaks every answer burns characters fast on a per-character meter. On a flat stream the agent talks as much as the work needs and nothing is counted.
- Standard tool shape, plugs into any tool list
- One stream carries a whole agent fleet
- No per-character meter on agent speech
Notes
Does the tool work with any LangChain agent?
Yes. It is a standard LangChain tool, so any agent that accepts a tool list can use it.
How do I set the voice?
Pass a voice id at construction, for example GandrText2SpeechTool(voice="gandr-leo"). The key is read from GANDR_API_KEY.