Transcribe is a tool for transcribing audio files using Google Speech API. It is intended for bulk processing of large (> 1 min) audio files -- such as from dictation recorders -- and automates GCS upload (and removal). It supports 44.1kHz .wav files only.
First, ensure you have Google Speech API enabled in your project as described here. Note that using the Google Speech API may not be free.
Second, install gcloud and allow application default credentials:
$ gcloud auth application-default login
Third, install 'sox' if stereo conversion is needed:
$ apt-get install sox
or equivalent. On OSX, an option would be $ brew install sox.
Fourth, install the transcribe tool:
$ go get github.com/herohde/transcribe
$ go install github.com/herohde/transcribe/cmd/transcribe
Then run:
$ transcribe --project=myproject [options] file [...]
By default, it will transcribe 'bar/foo.wav' into 'foo.wav.txt'. Add --mono if
stereo files.
Transcribe is released under the MIT License.