Voice recording made during different sessions are not the same volume. Is there a way to make the volume more equal or do I just have to start over?
The correct way to equalize perceived loudness across different 'takes' is to use a loudness meter. There are various free ones available, but you should be looking for a plugin that measures LUFS loudness.
These plugins are often known as R128 meters or LUFS meters and will usually measure in three different ways:
- Short-Term loudness
- Integrated Loudness
These tools are often used in post-production applications to measure the overall loudness of a piece of content so that it sits at the same loudness level with other content preceding or following it. It's typical use is to align television commercials in loudness by measuring integrated loudness, but has a wide application usage.
In order to equalize loudness in your application, measure the integrated loudness of each of your segments and adjust the level based on the meter output.
Another feature of loudness meters (driven by the loudness measurement standard) is the implementation of true peak metering. True peaks are audio peaks that appear in the analogue audio that may be greater in value then the surrounding digital audio sample peaks. It is perfectly possible to have two adjacent samples reading 0 dBFS and a true peak that exceeds 0dBTP - it will most likely clip the digital to analogue stage - even though the sample values are within the decoder range, so it is necessary to ensure that you never exceed 0dBTP.
Obviously it's better to nail this at the recording stage, by ensuring a consistent gain structure and distance to the microphone, but if you are 'in a bind' as it were, the correct way is to use loudness metering. Stay well away from normalization as you will end up chasing your tail because it will not give you an accurate result and will not work as expected.
In addition to Mark's excellent answer, there is potential in a multi-band compressor if your issue isn't only levels but also changing EQs due to mic position changes, on or off-axis recording, etc.
This doesn't come for 'free' - it will change the overall perceived 'shape' of the sound, and might work best if it's a vocal performance in a song as opposed to a voice-over or text reading etc but you could use it to level out such as consonants, 'boom' in the low-end of the voice if it was too close to the mic. It can even go some way towards masking mic pops.
This is a link to a Sound on Sound guide to using multi-band commpression in various situations. There's far too much to précis here.