Attempting to use telephone model, get error Upper frequency 6800.0 is higher than samprate/2 (4000.0)

[Eric Bunch]

Hello,

I'm trying to use pocketsphinx to transcribe some recorded phone calls, and running into some trouble, and am hoping for some help. I have an audio file out.wav that I beleive meets the format criteria: the output of sox --i out.wav is

Input File : 'out.wav' Channels : 1 Sample Rate : 8000 Precision : 16-bit Duration : 00:00:48.04 = 384320 samples ~ 3603 CDDA sectors File Size : 769k Bit Rate : 128k Sample Encoding: 16-bit Signed Integer PCM.

I am using pocketsphinx-python, with the following script

`
modeldir = "/Users/eric.bunch/psphinx/pocketsphinx/model"

config = Decoder.default_config()
config.set_string('-hmm', path.join(modeldir, 'en-us/en-us'))
config.set_string('-lm', path.join(modeldir, 'en-us/en-us-phone.lm.bin'))
config.set_string('-dict', path.join(modeldir, 'en-us/cmudict-en-us.dict'))
config.set_float("-samprate", 8000.0)

decoder = Decoder(config)
decoder.start_utt()

stream = open(path.join("/Users/eric.bunch/Downloads", 'out.wav'), 'rb')
while True:
buf = stream.read(4096)
if buf:
decoder.process_raw(buf, False, False)
else:
break
decoder.end_utt()
hypothesis = decoder.hyp()

print hypothesis.hypstr`

My current configuration is as follows:

[NAME] [DEFLT] [VALUE]
-agc none none
-agcthresh 2.0 2.000000e+00
-allphone
-allphone_ci no no
-alpha 0.97 9.700000e-01
-ascale 20.0 2.000000e+01
-aw 1 1
-backtrace no no
-beam 1e-48 1.000000e-48
-bestpath yes yes
-bestpathlw 9.5 9.500000e+00
-ceplen 13 13
-cmn current current
-cmninit 8.0 40,3,-1
-compallsen no no
-debug 0
-dict /Users/eric.bunch/psphinx/pocketsphinx/model/en-us/cmudict-en-us.dict
-dictcase no no
-dither no no
-doublebw no no
-ds 1 1
-fdict /Users/eric.bunch/psphinx/pocketsphinx/model/en-us/en-us/noisedict
-feat 1s_c_d_dd 1s_c_d_dd
-featparams /Users/eric.bunch/psphinx/pocketsphinx/model/en-us/en-us/feat.params
-fillprob 1e-8 1.000000e-08
-frate 100 100
-fsg
-fsgusealtpron yes yes
-fsgusefiller yes yes
-fwdflat yes yes
-fwdflatbeam 1e-64 1.000000e-64
-fwdflatefwid 4 4
-fwdflatlw 8.5 8.500000e+00
-fwdflatsfwin 25 25
-fwdflatwbeam 7e-29 7.000000e-29
-fwdtree yes yes
-hmm /Users/eric.bunch/psphinx/pocketsphinx/model/en-us/en-us
-input_endian little little
-jsgf
-keyphrase
-kws
-kws_delay 10 10
-kws_plp 1e-1 1.000000e-01
-kws_threshold 1 1.000000e+00
-latsize 5000 5000
-lda
-ldadim 0 0
-lifter 0 22
-lm /Users/eric.bunch/psphinx/pocketsphinx/model/en-us/en-us-phone.lm.bin
-lmctl
-lmname
-logbase 1.0001 1.000100e+00
-logfn
-logspec no no
-lowerf 133.33334 1.300000e+02
-lpbeam 1e-40 1.000000e-40
-lponlybeam 7e-29 7.000000e-29
-lw 6.5 6.500000e+00
-maxhmmpf 30000 30000
-maxwpf -1 -1
-mdef /Users/eric.bunch/psphinx/pocketsphinx/model/en-us/en-us/mdef
-mean /Users/eric.bunch/psphinx/pocketsphinx/model/en-us/en-us/means
-mfclogdir
-min_endfr 0 0
-mixw
-mixwfloor 0.0000001 1.000000e-07
-mllr
-mmap yes yes
-ncep 13 13
-nfft 512 512
-nfilt 40 25
-nwpen 1.0 1.000000e+00
-pbeam 1e-48 1.000000e-48
-pip 1.0 1.000000e+00
-pl_beam 1e-10 1.000000e-10
-pl_pbeam 1e-10 1.000000e-10
-pl_pip 1.0 1.000000e+00
-pl_weight 3.0 3.000000e+00
-pl_window 5 5
-rawlogdir
-remove_dc no no
-remove_noise yes yes
-remove_silence yes yes
-round_filters yes yes
-samprate 16000 8.000000e+03
-seed -1 -1
-sendump /Users/eric.bunch/psphinx/pocketsphinx/model/en-us/en-us/sendump
-senlogdir
-senmgau
-silprob 0.005 5.000000e-03
-smoothspec no no
-svspec 0-12/13-25/26-38
-tmat /Users/eric.bunch/psphinx/pocketsphinx/model/en-us/en-us/transition_matrices
-tmatfloor 0.0001 1.000000e-04
-topn 4 4
-topn_beam 0 0
-toprule
-transform legacy dct
-unit_area yes yes
-upperf 6855.4976 6.800000e+03
-uw 1.0 1.000000e+00
-vad_postspeech 50 50
-vad_prespeech 20 20
-vad_startspeech 10 10
-vad_threshold 2.0 2.000000e+00
-var /Users/eric.bunch/psphinx/pocketsphinx/model/en-us/en-us/variances
-varfloor 0.0001 1.000000e-04
-varnorm no no
-verbose no no
-warp_params
-warp_type inverse_linear inverse_linear
-wbeam 7e-29 7.000000e-29
-wip 0.65 6.500000e-01
-wlen 0.025625 2.562500e-02.

The error that I am receiving is

ERROR: "fe_interface.c", line 274: Upper frequency 6800.0 is higher than samprate/2 (4000.0).

Any help is much appreciated.

Best,
Eric

[Nickolay V. Shmyrev]

en-us provided with pocketsphinx is 16khz model. 8khz model is available in downloads, you need to use it instead.

[Eric Bunch]

Thank you, that's working. I do have a follow-up question because I think I may be misunderstanding something. My script is now

modeldir = "/Users/eric.bunch/psphinx/pocketsphinx/model"

config = Decoder.default_config()
config.set_string('-hmm', '/Users/eric.bunch/Downloads/en-us-8khz/')
config.set_string('-lm', path.join(modeldir, 'en-us/en-us.lm.bin'))
config.set_string('-dict', path.join(modeldir, 'en-us/cmudict-en-us.dict'))
config.set_float("-samprate", 8000.0)

decoder = Decoder(config)
decoder.start_utt()

stream = open(path.join("/Users/eric.bunch/Downloads", 'out.wav'), 'rb')
while True:
buf = stream.read(4096)
if buf:
decoder.process_raw(buf, False, False)
else:
break
decoder.end_utt()
hypothesis = decoder.hyp()

print hypothesis.hypstr

But it gives no output when I use the language model en-us-phone.lm.bin, which was unexpected for me. Is en-us-phone.lm.bin not the language model trained on phone conversations, or am I mistaken?

[Nickolay V. Shmyrev]

Is en-us-phone.lm.bin not the language model trained on phone conversations, or am I mistaken?

It is a model for phonetic recognizer.

[Eric Bunch]

Ah, ok. Thank you very much!