[Bug]: Multiple keyterm support for Listen v1 and v2 is broken in sdk v5.3

[konomith]

Summary

Unable to properly pass list of keyterms to connect method in sdk v5.3 for either api version

What happened?

Using deepgram-sdk==5.3.0 for listen endpoints v1 and v2 for both nova-3 and flux-general-en, we are unable to properly send a list of terms to the connect method.

Both client.listen.v1.connect and client.listen.v2.connect expect keyterm to be a str and passing a list won't append to the query_params properly:

• v1 relevant code
• v2 relevant code

As mentioned in this related (v2 only though) issue, this results in the SDK forming a url-encoded list of strings for the keyterm param, instead of properly adding each term separately. As a result, this breaks logic around passing multiple individual keyterms, and also makes it harder to stay within the 100 char limit per term, and the 500 tokens in total.

For v2, this issue was temporarily fixed with the v5.1.0 release but it seems to have regressed in the newer releases.
For v1, this has been broken since the v4 -> v5 change.

Steps to reproduce

1. Install deepgram-sdk==5.3.0
2. Try to pass multiple keyterms keyterm=["term1", ... "termN"] to either client.listen.v1.connect or client.listen.v2.connect
3. Observe that they are passed as a url-encoded list of strings instead of individually &keyterm=term1...&keyterm=termN

Minimal code sample

from deepgram import DeepgramClient

client = DeepgramClient(api_key="key")
terms = ["term1", "term2"]

# can also replace v1 with v2
with client.listen.v1.connect(keyterm=terms) as connection:
    # do something with the connection

Logs / traceback

# passing a lot of terms as a url-encoded list will break the length limit faster resulting in the following:

  File "/usr/local/lib/python3.11/site-packages/decorator.py", line 235, in fun
    return caller(func, *(extras + args), **kw)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.11/site-packages/retry/api.py", line 73, in retry_decorator
    return __retry_internal(partial(f, *args, **kwargs), exceptions, tries, delay, max_delay, backoff, jitter,
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.11/site-packages/retry/api.py", line 33, in __retry_internal
    return f()
           ^^^
  File "/app/service/plat/speech_to_text/deepgram.py", line 688, in _setup_deepgram_connection
    _dg_connection = self._deepgram_connection_exit_stack.enter_context(
                     ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.11/contextlib.py", line 517, in enter_context
    result = _enter(cm)
             ^^^^^^^^^^
  File "/usr/local/lib/python3.11/contextlib.py", line 137, in __enter__
    return next(self.gen)
           ^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.11/site-packages/deepgram/listen/v1/client.py", line 217, in connect
    with websockets_sync_client.connect(ws_url, additional_headers=headers) as protocol:
         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.11/site-packages/websockets/sync/client.py", line 308, in connect
    connection.handshake(
  File "/usr/local/lib/python3.11/site-packages/websockets/sync/client.py", line 95, in handshake
    raise self.protocol.handshake_exc
  File "/usr/local/lib/python3.11/site-packages/websockets/client.py", line 340, in parse
    self.process_response(response)
  File "/usr/local/lib/python3.11/site-packages/websockets/client.py", line 151, in process_response
    raise InvalidStatus(response)
websockets.exceptions.InvalidStatus: server rejected WebSocket connection: HTTP 400

Transport

WebSocket

API endpoint / path

/v1/listen and /v2/listen

Model(s) used

nova-3 for v1 and flux-general-en for v2

How often?

Always

Is this a regression?

• Yes, it worked in an earlier version

Last working SDK version (if known)

5.1.0 for v2, 4.8.1 for v1

SDK version

5.3.0

Python version

3.11.14

Install method

uv

OS

Linux (x86_64)

Environment details

Link to minimal repro (optional)

No response

Session ID (optional)

No response

Project ID (optional)

No response

Request ID (optional)

No response

Code of Conduct

• I agree to follow this project’s Code of Conduct