Skip to content

Open-source speech recognition toolkit for training, inference, streaming ASR, VAD, punctuation, speaker diarization pipelines, and OpenAI-compatible/MCP serving.

batch

游雁db89040b78c9
diff --git a/data/list/data_jsonl.list b/data/list/data_jsonl.list
new file mode 100644
index 000000000..d13808bf7
--- /dev/null
+++ b/data/list/data_jsonl.list
@@ -0,0 +1,2 @@
+/Users/zhifu/funasr1.0/data/list/train.jsonl
+/Users/zhifu/funasr1.0/data/list/val.jsonl
\ No newline at end of file
diff --git a/data_tmp/train.jsonl b/data_tmp/train.jsonl
new file mode 100644
index 000000000..e69de29bb
diff --git a/funasr/datasets/audio_datasets/update_jsonl.py b/funasr/datasets/audio_datasets/update_jsonl.py
index 561811e8e..0a3e0ec42 100644
--- a/funasr/datasets/audio_datasets/update_jsonl.py
+++ b/funasr/datasets/audio_datasets/update_jsonl.py
@@ -35,7 +35,7 @@ def gen_scp_from_jsonl(jsonl_file, jsonl_file_out, ncpu):
for line in lines:
- jsonl_file_out_f.write(line)
+ jsonl_file_out_f.write(line + "\n")
jsonl_file_out_f.flush()
jsonl_file_out_f.close()
@@ -53,6 +53,7 @@ def update_data(lines, i):
if (source_len_old - source_len) > 100 or (source_len - source_len_old) > 100:
print(f"old: {source_len_old}, new: {source_len}, wav: {wav_path}")
data["source_len"] = source_len
+ data["source"] = wav_path
jsonl_line = json.dumps(data, ensure_ascii=False)
lines[i] = jsonl_line