I would also do a quick syntactical/grammatical check.
When you read an actual transcript made by "reversing a tape" you will often find a lot of "<missing>""<not understandable> or [some word?] etc., the actual sentences will not be all "properly" formulated, some will have missing subjects, or verbs, some "slang" and consider how most of them are "human-interpreted" (by an experienced professional) and corrected for syntax/grammar and "sense".
An automated machine, especially if recording "room conversations" through a tiny mic of a laptop would have several garbled sentences, people don't normally talk clearly and in the direction of the microphone even when asked to, and if you ever used a transcription/voice activated software (Dragon Naturally Speaking to name one) you would know how clearly you need to speak (and which kind of "training" you have to do with the software) before having satisfying result).
So, such a software may well exist, but IMHO the result it would provide would be - for a recording in the conditions described - of extremely poor quality, with a lot of garbage, gaps and what not.
Additionally, at least all the tools of similar nature I have seen have as "input" an audio file (or stream through the mic) and output some form of text.
Of course you can then feed the text transcript to a "computer reader" and record the .wav, but it would make little sense, unless it is for someone who is visually impaired.
What you describe sounds more like a subtitled audio (or video) recording, there are tools, but they all have the same accurateness issues mentioned above, see (example):
https://wiki.gnome.org/Gaupol/SpeechRecognition
Of the Commercial softwares for Mac one of the best is considered this one:
http://www.nuance.com/for-individuals/by-product/dragon-for-mac/macspeech-scribe/index.htm
but like most others it is more about "dictation", and:
Quote::
Transcribed audio must be of a single speaker's dictation in one of these file formats: .wav, .aif, .aiff, .m4v, .mp4, or .m4a
jaclaz
↧