Fix for crash in alignment when called with no tokens

[balooii]

Fixes #1048

There are probably better ways to do this but this change restores previous behavior which worked before changes from 65b2332
-> when called with empty list of tokens return empty tensor

[ZodiacFRA]

Worked for me, thanks

EDIT: I was using version 3.3.1

[SoftwareAndOutsourcing]

I haven't had this issue since upgrading to v3.3.2. Is it still necessary to address it?

[ZodiacFRA]

Hi @SoftwareAndOutsourcing

I was using version 3.3.1, I did not try with 3.3.2.

However in alignment.py, get_wildcard_emission() is the same in both versions so it might be worth it to add this fix (unless the bug has been fixed in another part of the code)

Thank you for your work anyway :)
Have a nice day

[uduntuntu]

I patched that issue for upper boundary also. Test code below

# Run (in virtualenv):  python test_alignment_out_of_bounds.py
import torch
from whisperx import alignment as A

def run_case(V, tokens, blank_id=0, label=""):
    fe = torch.randn(V, dtype=torch.float32)
    t = torch.tensor(tokens, dtype=torch.long)
    try:
        out = A.get_wildcard_emission(fe, t, blank_id)
        ok = bool(torch.isfinite(out).all())
        print(f"[OK ] {label:40s} -> out.shape={tuple(out.shape)} finite={ok}")
    except Exception as e:
        print(f"[ERR] {label:40s} -> {type(e).__name__}: {e}")

print("Testing whisperx.alignment.get_wildcard_emission")
print("Function:", A.get_wildcard_emission.__code__.co_filename)

# OFF-BY-ONE repro: token == vocab size (V)
run_case(34, [34], 0, "off-by-one (V=34, token=V)")

# Mixed case: wildcard + valid + token==V
run_case(34, [-1, 0, 5, 34], 0, "wildcard + in-range + token=V")

# Edge case: empty emissions (should assert / error, but not IndexError from indexing)
run_case(0, [0], 0, "empty emissions (V=0)")