diff --git a/wavinfo/umid_parser.py b/wavinfo/umid_parser.py
index faa511c..fcbddda 100644
--- a/wavinfo/umid_parser.py
+++ b/wavinfo/umid_parser.py
@@ -1,5 +1,3 @@
-from typing import Union
-import binascii
 from functools import reduce
 
 
@@ -13,7 +11,7 @@ class UMIDParser:
 
     This implementation is based on SMPTE ST 330:2011
     """
-    def __init__(self, raw_umid: bytearray):
+    def __init__(self, raw_umid: bytes):
         self.raw_umid = raw_umid
     #
     # @property
diff --git a/wavinfo/wave_bext_reader.py b/wavinfo/wave_bext_reader.py
index fdbc230..a2a9561 100644
--- a/wavinfo/wave_bext_reader.py
+++ b/wavinfo/wave_bext_reader.py
@@ -1,7 +1,7 @@
 import struct
-import binascii
 from .umid_parser import UMIDParser
 
+from typing import Optional
 
 class WavBextReader:
     def __init__(self, bext_data, encoding):
@@ -16,44 +16,44 @@ class WavBextReader:
         rest_starts = struct.calcsize(packstring)
         unpacked = struct.unpack(packstring, bext_data[:rest_starts])
 
-        def sanitize_bytes(b):
+        def sanitize_bytes(b : bytes) -> str:
             first_null = next((index for index, byte in enumerate(b) if byte == 0), None)
             trimmed = b if first_null is None else b[:first_null]
             decoded = trimmed.decode(encoding)
             return decoded
 
         #: Description. A free-text field up to 256 characters long.
-        self.description = sanitize_bytes(unpacked[0])
+        self.description : str = sanitize_bytes(unpacked[0])
         #: Originator. Usually the name of the encoding application, sometimes
         #: a artist name.
-        self.originator = sanitize_bytes(unpacked[1])
+        self.originator : str = sanitize_bytes(unpacked[1])
         #: A unique identifier for the file, a serial number.
-        self.originator_ref = sanitize_bytes(unpacked[2])
+        self.originator_ref : str = sanitize_bytes(unpacked[2])
         #: Date of the recording, in the format YYY-MM-DD
-        self.originator_date = sanitize_bytes(unpacked[3])
+        self.originator_date : str = sanitize_bytes(unpacked[3])
         #: Time of the recording, in the format HH:MM:SS.
-        self.originator_time = sanitize_bytes(unpacked[4])
+        self.originator_time : str = sanitize_bytes(unpacked[4])
         #: The sample offset of the start of the file relative to an
         #: epoch, usually midnight the day of the recording. 
-        self.time_reference = unpacked[5]
+        self.time_reference : int  = unpacked[5]
         #: A variable-length text field containing a list of processes and
         #: and conversions performed on the file.
-        self.coding_history = sanitize_bytes(bext_data[rest_starts:])
+        self.coding_history : str = sanitize_bytes(bext_data[rest_starts:])
         #: BEXT version. 
-        self.version = unpacked[6]
+        self.version : int = unpacked[6]
         #: SMPTE 330M UMID of this audio file, 64 bytes are allocated though the UMID
         #: may only be 32 bytes long.
-        self.umid = None
+        self.umid : Optional[bytes] = None
         #: EBU R128 Integrated loudness, in LUFS.
-        self.loudness_value = None
+        self.loudness_value : Optional[float] = None
         #: EBU R128 Loudness rante, in LUFS.
-        self.loudness_range = None
+        self.loudness_range : Optional[float] = None
         #: True peak level, in dBFS TP
-        self.max_true_peak = None
+        self.max_true_peak : Optional[float] = None
         #: EBU R128 Maximum momentary loudness, in LUFS
-        self.max_momentary_loudness = None
+        self.max_momentary_loudness : Optional[float] = None
         #: EBU R128 Maximum short-term loudness, in LUFS.
-        self.max_shortterm_loudness = None
+        self.max_shortterm_loudness : Optional[float] = None
 
         if self.version > 0:
             self.umid = unpacked[7]