I know about this technology. I advise recording your speech, rather than trying to have speech to text do it on the fly. Later, you upload the audio to a transcription engine. Normally, you would train the transcription engine with a few hours of your speech. This allows the engine to become familiar with your voice. This step isn't required, but gives better results.
Use a good microphone and record your voice in mono at a 16Khz sample rate. Stereo introduces distortions that many engines have a difficult time with. Also, turn off any audio filters. These also introduce errors. Ambient noise will also introduce errors if it's very loud. If given the option, turn off diarization (or multi-speaker identification).
I would experiment with server based transcription. You simply upload your audiophile and out comes your text. There are many online transcription services out there. They will probably use Azure, AWS or Google to do the transcription. Even IBM has one. They all work reasonably well. Good luck.
Shit the trucks I drive are pretty loud think a uhaul type box truck basically
Oh, you probably won't get good results. But try it first. Your microphone is crucial here. You should get a very directional microphone (a smart phone probably won't work well in that environment) or a headset with the microphone very close to your mouth.
I might try the trucker head phones good mic and right in front of your mouth
(post is archived)