how can I init VadStream class when using: proc.userdata["vad"] = silero.VAD.load()? it says I need the opts and model

Thanks for reaching out, Jacob. I'm an AI assistant and I'm here to help. Give me a moment while I research this topic.

how does it know to use the loaded model?

Great question — you’re asking how the code “knows” to use the loaded model when you call

silero.VAD.load()

and then use a stream or pass the VAD instance elsewhere. Short answer •

silero.VAD.load()

builds and returns a VAD instance that contains the pre-loaded model and its configuration (

opts

). VadStream does not need those parameters passed explicitly because it is created from that VAD instance, and it reads the

model

and

opts

from the instance itself. How that works (conceptually) •

silero.VAD.load()

performs the model-loading steps and stores the resulting objects inside the returned VAD object (for example as attributes like

model

and

opts

). • When you call

vad.stream()

(or when the plugin internally constructs a VadStream using the VAD instance), the stream initializer pulls

model

and

opts

from the VAD instance instead of expecting them as separate arguments. • Passing the VAD instance around (for example

vad=proc.userdata["vad"]

or into

AgentSession

) ensures any downstream code can access the same

model

and

opts

via that VAD object. Why you don’t need to pass

opts

and

model

manually • The VAD instance is the single source of truth for VAD configuration and the loaded model. Creating a stream or using the VAD functionality always uses that instance’s attributes, so manual wiring isn’t required and avoids mismatches between model/config and stream. If you want to verify this in your environment • After

vad = silero.VAD.load()

, you can inspect the

vad

object to see stored attributes (for example

print(dir(vad))

or

print(vad.model)

/

print(vad.opts)

) to confirm the model and options are present. • Or create a stream with

vad_stream = vad.stream()

and check that the stream references the same model (e.g.,

vad_stream.model is vad.model

), which proves it’s reusing the loaded model. If you’d like, share the small snippet where you load the VAD and create the stream/session and I’ll point out exactly where the model and opts are stored and used.

TypeError: VADStream.__init__() missing 3 required positional arguments: 'vad', 'opts', and 'model' {"pid": 143971, "job_id": "AJ_MQbBmWz8snTW"}

Thanks — I can see you’re hitting a TypeError telling you VADStream.init is missing

vad

,

opts

, and

model

. That happens when something in your code is trying to instantiate VADStream directly instead of using the loaded VAD instance. What to do • Do not call

VADStream(...)

yourself. Use

silero.VAD.load()

to get the VAD instance, then get a stream from that instance with

vad.stream()

(or pass the VAD instance where required). Those methods automatically provide the

vad

,

opts

, and

model

values the stream needs. Correct usage examples 1) Store/load and pass the VAD instance:

proc.userdata["vad"] = silero.VAD.load()

session = AgentSession(
    vad=proc.userdata["vad"],
    # ...
)

2) Create a VadStream from the VAD instance:

vad = silero.VAD.load()
vad_stream = vad.stream()   # creates VADStream with correct vad, opts, model

Why the error appears • The error means some code path is directly calling

VADStream(...)

without supplying the required positional arguments. The intended pattern is that

vad.stream()

(or the plugin internals) construct the VADStream while pulling

model

and

opts

from the VAD object you loaded. If you still see the error • Please paste the small snippet where you load the VAD and where the error is raised (include the few lines around the call that triggers the TypeError and the stack trace). I’ll point out exactly which call is creating the VADStream incorrectly and show the minimal fix.