Text-to-Speech / inference.sh
zyingt's picture
Upload 685 files
0d80816
raw
history blame
1.3 kB
#!/bin/bash
Check if a text file is provided as an argument
if [ "$#" -ne 1 ]; then
echo "Usage: $0 <text_file>"
exit 1
fi
text_file="$1"
# Check if the provided text file exists
if [ ! -f "$text_file" ]; then
echo "Error: Text file '$text_file' not found."
exit 1
fi
# Define the list of speaker IDs
speakers="92 6097 6670 6671 8051 9017 9136 11614 11697 12787"
counter=1
# Read lines from text_file.txt
while IFS= read -r line; do
# Skip empty lines
[ -z "$line" ] && continue
label="text$counter"
counter=$((counter+1))
# Iterate through each fixed speaker ID
for speaker_id in $speakers; do
output_dir="/mnt/workspace/tzeying/inference/inf_librihifi_320k/trial1/${label}/${speaker_id}"
speaker_name="hifitts_${speaker_id}"
sh /mnt/workspace/wangmingxuan/vits_on_libritts_hifitts/Amphion/egs/tts/vits_hifitts/run.sh \
--stage 3 \
--gpu "4" \
--infer_expt_dir /mnt/workspace/xueliumeng/data/vits_on_libritts_hifitts/logs/Libri_HifiTTS_All \
--infer_output_dir "$output_dir" \
--infer_mode "single" \
--infer_text "$line" \
--infer_speaker_name "$speaker_name"
echo "Processing complete for speaker ID: $speaker_id"
done
echo "Processing complete for line: $counter"
done < "$text_file"